AN ABSTRACT OF THE THESIS OF
Irene D. Dumkow for the degree of Doctor of Philosophy in Physics presented
on December 1, 1997. Title: Oxygendeficient YBaSu30_
Films Investigated
by Perturbed Angular Correlation Spectroscopy.
Redacted for Privacy
Abstract approved:
Janet Tate
Perturbed angular correlation spectroscopy (PAC) is well suited to study
phase transitions in materials because its sensitivity is temperature
independent, unlike that of the related nuclear hyperfine techniques of
nuclear quadrupole resonance and Mogbauer spectroscopy. PAC was used to
study the normal state of YBa2Cu306,, a high temperature superconductor
which shows unique and interesting features with changing oxygen content.
nrrn, which substitutes for Y in YBa2Cu306,, was the PAC probe.
In this thesis, the direction of the main axis of the electric field gradient
tensor at the Y site in YBa2Cu306, was determined. It points along the chain
direction or b-axis of fully oxygenated YBa2Cu3O7. To establish this result, a-
axis oriented films were grown and studied by PAC, and the growth mode
characterized.
This work also used PAC to establish the origin of a site in some PAC
frequency spectra at vc2 = 133 MHz, ri = 0. This site appears only under strongly
reducing conditions. It arises from CuYO2 in the (012) orientation and
information about the conditions under which this material forms was
gathered, especially with respect to the temperature and pressure. This
material, along with BaCu2O2 in the (112) orientation, was also observed in xray diffraction.
In order to investigate films with varying oxygen content, a system was
developed and tested that enabled the oxygen content to be changed reliably.
PAC experiments were performed on films with oxygen contents between
6.25 and 7 and it was found that the nuclear quadrupole frequency decreases
with decreasing temperature, and that the asymmetry has a maximum for a
nominal oxygen content of 6.7. This may be evidence of oxygen ordering.
Strong relaxation is observed at all temperatures and this may be a
characteristic of the probe rather than the host material.
©Copyright by Irene D. Dumkow
December 1, 1997
All Rights Reserved
Oxygendeficient YBa2Cu306, Films Investigated by Perturbed Angular
Correlation Spectroscopy
by
Irene D. Dumkow
A THESIS
submitted to
Oregon State University
in partial fulfillment of
the requirement for the
degree of
Doctor of Philosophy
Presented December 1, 1997
Commencement June 1998
Doctor of Philosophy thesis of Irene D. Dumkow presented on December 1,
1997
APPROVED:
Redacted for Privacy
Major Professor, representing Physics
Redacted for Privacy
Chair of Department of Physics
Redacted for Privacy
Dean of Gr
uate School
I understand that my thesis will become part of the permanent collection of
Oregon State University libraries. My signature below authorizes release of
my thesis to any reader upon request.
Redacted for Privacy
Irene D. Dumkow, Author
Acknowledgement
I would like to acknowledge the Baden-Wiirttemberg
Oregon
exchange program, without which it would not have been possible for me to
study at Oregon State University.
My parents supported me financially as well as emotionally by firmly
believing that I can reach what I have set out to accomplish.
I have to thank Dr. Janet Tate for all the guidance and support received
during the last couple of years. Without her and her door which was always
open, this thesis would not exist today.
I have to thank Dr. Roland Platzer for all his expertise in carrying out
the PAC measurements. Without him, this thesis would not have been
possible.
I have to acknowledge the support of Dr. John Gardner. It was in his
laboratories that we took the PAC measurements and he was always there
with helpful discussion and suggestions for our data.
I have to thank Dr. Dennis Tom for "showing me the ropes" and
having done all the work in developing the evaporation system. I always was
grateful about his prompt reply to email questions.
I want to extend my thanks to Dr. Jeanette Roberts and Dr. Goran
Karapetrov. They made me feel welcome in the research group and were
always willing to answer my questions.
My special thanks go to Dr. Brandon Brown, who always had an open
ear to answer questions.
I also want to thank Amy Droegemeier, Eric Moret and Andrew
Draeseke , who as the new "crop" will carry on the work in the lab and lifted
my spirits quite often.
I would like to thank Dr. A. Sleight for the use of his
x-ray
diffractometer.
Dr. Roberta Bigelow has been very helpful in performing additional xray measurements.
I would like to thank Dr. C. Thomsen for performing the Raman
measurement, Dr. L. McIntyre for the RBS measurement and Dr. R. Nielsen
for the EMPA measurements.
The financial supports from the Physics Department and the National
Science Foundation under grant number DMR 9704052 are gratefully
acknowledged.
Table of Contents
Page
1. INTRODUCTION
1
1.1 Structure of YBa2Cu306,
2
1.2 Differences Between Crystals and Films
7
1.3 Examples of Differences between Films and Crystals
8
1.3.1 Tc versus Oxygen Content
1.3.2 c-Axis Length versus Oxygen Content
1.4 Role of Oxygen
1.4.1 Chain Length and Charge Transfer
1.4.2 Oxygen Ordering: Experiment and Theory
2. SAMPLE PREPARATION
10
11
12
13
13
17
2.1 Thin Film Deposition Techniques
17
2.2 Reactive Co-Evaporation
18
2.3 Factors Influencing the Film Growth
20
2.3.1 Stoichiometry
2.3.2 Substrate Temperature During the Deposition
2.3.3 Oxygen Pressure During Deposition
2.3.4 Postanneal: Oxygen Pressure
2.3.5 Postanneal: Substrate Temperature
20
20
21
21
21
2.4 Description of the Evaporation System
22
2.5 Control of Oxygen Content
23
2.5.1 Description of Apparatus
2.5.2 General Description of the Procedure
24
26
Table of Contents (continued)
Page
3. CHARACTERIZATION
29
3.1 Raman Spectroscopy
29
3.2 Rutherford Backscattering
33
3.3 The Transition Temperature
37
3.4 X-Ray Diffraction
41
3.4.1 Interpretation of X-Ray Data
3.4.2 Determination of Lattice Parameter from X-Ray Data
4. PERTURBED ANGULAR CORRELATION SPECTROSCOPY
50
56
58
4.1 Naive Theory
59
4.2 In-Depth Treatment
62
4.3 General Perturbation Function
64
4.4 Data Acquisition
71
4.5 Data Analysis
73
4.6 Perturbed Angular Correlation Experiments in YBa2Cu307.6
75
5. aAXIS ORIENTED YBA2CU3O6+X FILMS
80
5.1 X-Ray and Resistance Studies on a-Axis Oriented Films
85
5.2 Investigation of a-Axis Oriented Films with PAC
94
Table of Contents (continued)
Page
6. OXYGEN-DEFICIENT FILMS
98
6.1 Investigation of Tetragonal YBa2Cu306,
98
6.2 Previous Work on Tetragonal YBa2Cu306+x Films
99
6.3 Starting Point for the Investigation
100
6.4 X-ray Studies on Tetragonal Films
103
6.5 Identification of the Second Site
105
6.6 OxygenDeficient YBa2Cu306,x : Examples from Literature
108
6.7 Early Results: Anneals in Evaporator
110
6.8 Results from the New System
113
6.9 PAC Results
119
7. SUMMARY
126
7.1 The Electric Field Gradient in YBa2Cu307_5 Films
126
7.2 Impurity Phases in YBa2Cu306,
127
7.3 Oxygendeficient Films
129
7.4 Future Direction
130
BIBLIOGRAPHY
APPENDICES
132
139
List of Figures
Page
The structure of an ideal perovskite. They can either be cubic,
tetragonal or orthorhomic, e.g. BaTiO3 can be all three of them,
depending on the temperature.
3
Sketches of the fully oxygenated YBa2Cu3O7 (a) structure and the
oxygen-deficient YBa2Cu3O6 (b) structure.
4
1.3
Sketch showing the buckling of the Cu-0 planes.
5
1.4
Tc vs oxygen content for crystals and films.
10
1.5
Definition of the potentials between oxygen and copper atoms.
14
1.6
Examples of the ordering of oxygen atoms in the basal plane for
an oxygen content of 6.5.
15
2.1
Phase diagram used to change the oxygen content of our films.
23
2.2
System used to control the oxygen content
25
3.1
Raman spectra taken on an orthorhombic film and two
tetragonal films subjected to different anneals.
32
3.2
Typical Rutherford Backscattering setup.
33
3.3
Sample histogram of a thin self-supporting film
35
3.4
Example of a R(T) -curve, where contacts became faulty
around 90 K.
40
3.5
X-ray diffraction in 0-20 mode (Bragg-Brentano geometry).
42
3.6
The path difference between rays scattered from two points of a
)
Bravais lattice is d (fi
43
Influence of the film thickness on the broadening of diffraction
peaks at different angles.
46
Influence of a variation in the lattice parameter on the peak width.
46
1.1
1.2
3.7
3.8
List of Figures (continued)
Eig,
3.9
3.10
3.11
Page
Schematic diagram of Field-Merchant method to determine in-
plane alignment of textured samples.
49
Overloading of the dectector due to intensely diffracting substrate
peaks.
53
Influence of the Cu KR line due to the MgO substrate on the shape
of the YBa2Cu3O7 peak.
53
3.12
9-20 spectra on the same YBa2Cu3O7 film deposited on MgO.
54
4.1
Decay scheme for mIn, showing only the transitions relevant
for PAC.
58
4.2
Emitted radiation pattern for 1= 1 and m = 0,±1 (dipole radiation).
59
4.3
Hypothetical decay scheme of a 0-1-0 cascade.
60
4.4
Definitions of angles and orientations for detector-probe system.
66
4.5
Definition of angles used to describe the orientation of the film
relative to the detectors.
67
Calculated amplitudes for different orientations, with the electric
field gradient perpendicular to the film plane or in the film plane.
69
4.6
4.7
R(t) and Fourier transform of a tetragonal (left) and an
orthorhombic film, with the films in the same orientation
(a = 45°,(3 = 90°).
5.1
70
Rise in the (200) peak of YBa2Cu3O7 with decreasing deposition
temperature.
86
5.2
Ratio of the maximum peak intensities of the (200) and (005) peaks. 88
5.3
Rocking curves of an a-axis YBa2Cu306+x film on a LaA103
substrate and the LaA1O3 alone.
88
List of Figures (continued)
Page
Eig,
5.4
Ratios of the integrated intensities of rocking curves on the (200)
and (005) peaks.
90
Dependence of the ratio of the integrated intensities of the (200)
and (005) peak (calculated from rocking curves) of YBa2Cu3O7 on
the pressure during deposition.
90
Normalized resistance for several films grown at decreasing
deposition temperatures with increasing a-axis content.
92
5.7
Development of T, and AT, with decreasing substrate temperature.
93
5.8
Dependence of T, on the oxygen pressure for films deposited
5.5
5.6
at 600 °C.
5.9
R(t) and Fourier transforms of an orthorhombic a-axis oriented
YBa2Cu3O7 film at room temperature.
5.10
6.3
6.4
6.5
6.6
95
R(t) and Fourier transform of two tetragonal films annealed at
500 °C and 550 °C.
6.2
94
R(t) and Fourier transform of a tetragonal a-axis oriented
YBa2Cu306.25 film.
6.1
93
102
X-ray spectra of YBa2Cu30625 films showing the two impurity
peaks near the (004) of YBa2Cu306.25.
104
Development of the T, and AT, for films annealed at different
pressures, if the pressure is adjusted only once.
110
Development of T, and AT, if the pressure is kept constant
during the cool-down and the anneal.
111
Resistance of two films annealed on the 6.6 phase line, one with
bulk YBa2Cu3O7, the other with no bulk material in the vicinity.
114
Change in c-axis length for PAC and non-PAC films and
comparison to published values.
115
List of Figures (continued)
Page
Figs
6.7
6.8
6.9
6.10
Tc for our oxygendeficient YBa2Cu306, films compared to
published data.
116
R(t) curves and Fourier transforms of all of the spectra taken
on HTC 34, with the anneals indicated.
122
Asymmetry and linewidth as a function of oxygen content for
HTC 34.
123
R(t) and Fourier transforms for HTC 35, annealed insitu at
different temperatures to achieve a 6.7 anneal.
125
List of Tables
Page
Table
3.1
Raman frequencies (in cm 1) for YBa2Cu3O7.
29
3.2
The expected 9 -20 values for c-axis oriented YBa2Cu306,.
51
3.3
Expected 20 values in ° for the substrates used in this thesis.
51
5.4.
20 (in °) values for LaA1O3 and a-axis and b-axis oriented
YBa2Cu3O7.
6.5
Frequency and asymmetry of the A and B sites.
6.6
20 values (in °) and the calculated d-values (in A) of the impurity
87
101
peaks for tetragonal YBa2Cu306.25 films subjected to different
post-deposition in-situ anneals.
6.7
6.8
7.9
105
Summary of anneals performed on several PACfilms, as well
as Tc and c-axis length, when applicable.
119
Oca /co for film HTC 34 annealed for nominal oxygen
contents of 7 (as deposited), 6.7, 6.5, 6.4 and 6.25.
121
vQ,
Summary of electriC field gradient in YBa2Cu3O7 and YBa2Cu306.25
at room temperature.
127
List of Appendices
Appendix
A. HOW TO GROW A FILM
Page
140
A.1 Choice and Cleaning of the Substrate
140
A.2 Loading the Chamber
141
A.3 Heating the Substrate before Deposition
142
A.4 Depositing the Film
143
A.5 Annealing the Film
144
A.6 Cool-down after Deposition
145
B. SETTINGS OF X-RAY DIFFRACTOMETER
146
C. FITTING OF X-RAY DATA
147
Oxygen-deficient YBa2Cu306, Films Investigated by Perturbed
Angular Correlation Spectroscopy
Chapter 1
Introduction
High Tc superconductors,
of which
YBa2Cu306, is the most
exhaustively studied, remain very intriguing materials despite 10 years of
intensive attention by researchers. The superconducting and normal
properties, as well as the crystalline structure, are extremely sensitive to the
oxygen content which varies from x=0 to x=1. Depending on the oxygen
content, the material either shows superconducting or semiconducting
behavior and the structure changes from tetragonal for low oxygen contents
to orthorhombic for high oxygen content. The materials are also extremely
anisotropic; their properties are different when measured parallel and
perpendicular to their trademark Cu-0 planes. For the resistivity, this
anisotropy ranges from pc/pab -405 for Bi2,Sr2_yCu06,6 [1] to o /oab
for
YBa2Cu307.5 1, one of the more isotropic high Tcmaterials.
The most reproducible information for the YBa2Cu306+ system has
come from the single crystal material. While much of this is applicable to
both thinfilm and singlecrystal forms of the materials, there remain
important differences between them, mostly having to do with processing
and disorder issues. These are of crucial importance for applications,
particularly those which require films
in
different
crystallographic
I Both YBa2Cu306+x and YBa2Cu307_8 are used in this thesis. YBa2Cu306+
usually, but not always, implies that one is looking at oxygendeficient
material, while YBa2Cu307_8 is most commonly used for optimally prepared
material.
2
orientations. This thesis focuses on thinfilm YBa2Cu306,, and makes use of
a unique thinfilm facility capable of introducing radioactive isotopes into
thinfilm superconductors during the growth process. This allows the use of
the hyperfine technique, perturbed angular correlation spectroscopy (PAC), to
investigate features of the structure of YBa2Cu306.x, as well as to identify
impurity phases. The remainder of chapter
1
discusses some general
properties of YBa2Cu307_8 and takes a brief look at the importance of the
oxygen content. Chapter 2 discusses the growth of the films and how to
change the oxygen content, while chapter 3 focuses on the characterization of
the films. Chapter 4 is a review of the work to date on PAC in YBa2Cu306..,
and describes the experimental set up and data analysis. Chapters 5 and 6
present the main experimental results of this thesis
X-ray and PAC
investigations of both a- and c-axis oriented YBa2Cu306 films with varying
oxygen content. Chapter 7 summarizes the results and includes prospects for
future investigations.
1.1 Structure of YBa2Cu306..
Superconductivity has been subject of research since 1911, when H.
Kamerlingh Orines discovered that mercury became superconducting below
4.15K [2]. The era of hightemperature superconductivity was initiated in
1986, when Muller and Bednorz first published their results [3] of
superconductivity at 36 K in a lanthanum barium copper oxide. After
confirmation of this discovery, which was especially surprising because it was
believed at that time that superconductivity could not exist above 30 K, a
flurry of research activity started, and intensified when it was discovered that
the yttriumbarium-copper-oxygen system became superconducting at = 90 K
3
[4], replacing the need for liquid helium as a cooling medium by liquid
nitrogen, which is cheaper and easier to handle. From the start, there has
been interest in growing these materials as thin films. This is due to the fact
that thinfilm superconductors can be used for devices and the potential to
operate superconducting devices at the commercially viable temperature of 77
K is an exciting prospect. Nowadays, process issues have been resolved and
commercially produced devices are available.
YBa2Cu306, structure is based on the perovskite structure. A true
perovskite has the formula ABO3. Fig. 1.1 shows the structure, which is a body
centered cube, the A atoms occupying the corners of the cube, while the B
atoms occupy the body-centers. The oxygen atoms are on each of the edges of
the cube.
O
Cation A
Cation B
O
Oxygen
Fig. 1.1 The structure of an ideal perovskite. They can either be cubic,
tetragonal or orthorhomic, e.g. BaTiO3 can be all three of them, depending on
the temperature.
4
The YBa2Cu3O6, structure has three such cubes, each of which is
deficient in some of the oxygen atoms. The Cu atoms occupy the A site, while
the position of the B site is taken up by either the Ba or Y. In the Y cube, the
oxygen atoms at the same height as the Y atoms are missing, while the lower
and upper oxygen-sites are fully occupied. This leaves the Y atom sandwiched
between two Cu-0 sheets or
planes,
which are the site
of the
superconductivity. In the case of the Ba cube, the picture is more complicated.
The oxygen atoms at the height of the Ba are there, as are all the oxygen atoms
at the side it shares with the Y cube. At the side that is left, there is great
(b)
(a)
Cu(2)
0(2)
Ba
0(4)
etPh 0(1)
Cu(1 )
Ba
0(4)
0(1)
Cu(1)
1111
Fig. 1.2 Sketches of the fully oxygenated YBa2Cu30, (a) structure and the
oxygen-deficient YBa2Cu3O6 (b) structure. Thermal vibration ellipsoids are
shown for the atoms. In the orthorhombic YBa2Cu307 structure the oxygen
0(1) atoms are arranged in long chains along the b-axis, in the tetragonal
YBa2Cu3O6 structure, the 0(1) oxygen atoms are randomly distributed in the
basal plane. (Plots taken from [5])
5
variability in the number of oxygen atoms, with consequences for the
structure and the superconductivity, which will be shown later in this thesis.
Fig. 1.2 shows the structure of both the fully oxygenated YBa2Cu307 and the
oxygen-deficient YBa2Cu3O6 and comparing this to the ideal perovskite
structure helps to see the differences.
In the case of fully oxygenated YBa2Cu306, (x--4), two of the possible
four sites are occupied and the oxygen atoms form long chains along the b-
direction in conjunction with the Cu atoms. These chains give the material
an overall orthorhombic structure, with ideal values for a=3.82A, b=3.89A
and c=11.68A [5]. The more oxygen is taken out, the more the a- and b-axis
/
A
a
Fig. 1.3 Sketch showing the buckling of the Cu-0 planes. Due to symmetry,
the Cu-0 chains are perfectly straight (Sketch from [6])
6
become equivalent, with an orthorhombictotetragonal transition at
x
0.35. In the fully oxygenated YBa2Cu3O7, the Cu-0 planes are buckled, see Fig.
1.3, but with decreasing oxygen content, the planes straighten, which explains
why the c-axis lattice parameter increases with decreasing oxygen content.
This thesis is concerned mostly with the structure of the material and
approaches it from several different points of view. Perturbed angular
correlation spectroscopy (PAC), a hyperfine technique which is based on the
interaction of the quadrupole moment of a probe atom with the distribution
of charges around said atom, can be used to measure the orientation of the
electric field gradient, which is a tensor quantity. Yet our films are twinned, as
a consequence of the way they grow, and at the start of the investigation we
only knew that the main axis of the electric field gradient lies in the ab-plane.
The first goal was therefore to grow films with the a-axis normal to the
substrate as opposed to the films we grew before, which had the c-axis normal
to the substrate.
After being able to grow the films with different orientations, we are
now trying to take a closer look at the role of oxygen in the material. This
interest was sparked by the fact that it is relatively easy to grow YBa2Cu306,
with an oxygen content of x = 7 in the orthorhombic structure or of x = 6.25 in
the tetragonal structure. In order to vary the oxygen content more, we first
had to learn as much as possible about what happens during the growth and
anneal right after the deposition. After investigating options of changing the
oxygen content right after the deposition, which did not prove to be very
reliable, we turned to building a system to change the oxygen content in a
post-deposition anneal outside the main evaporation chamber, which gives
us more control over the whole process.
7
1.2 Differences Between Crystals and Films
Even though crystals and thin films of YBa2Cu306, are made of the
same material, often they seem to be two very different beasts. This is already
evident in the preparation methods, which make the films more susceptible
to small changes in preparation. While papers dealing with crystals seem to
have reached some consensus about the properties of YBa2Cu306,, this
consensus has not quite been reached for films, making them at the same
time an interesting and challenging subject to study.
To see the difference between crystals and films, it is instructive to look
at the methods of preparation and the time-frames involved. In order to grow
good crystalline material, materials are usually mixed, sintered, ground and
reground and annealed several times in flowing oxygen. Jorgensen et al. [5],
for example, mixed powder of Y203, BaCO3 and CuO in the right proportions,
after they were individually prescreened through 200 mesh. This mixture was
pressed into pellets and put into a platinum crucible. The pellets were heated
to 960 °C in flowing oxygen and annealed for 24 h. Afterwards they were
cooled in air, ground to 200 mesh, remixed and pressed again into pellets.
Again they were taken up to 960 °C in flowing oxygen and kept there for 24 h.
Afterward the temperature was reduced to 670 °C and held for 24 h followed
by a cool-down to ambient temperature at 5°C/min. The sample spends three
days at a constant temperature and that does not include heating and cooling.
For single crystals, the annealing times are often several weeks in order to
completely oxygenate the crystal [7].
Compared to this elaborate and time-consuming procedure, making a
films of YBa2Cu306, is fast. There are several different methods commonly
used to deposit films, the most widely used being sputtering, laser ablation,
8
evaporation techniques like MBE (molecular beam epitaxy) and reactive co­
evaporation, and chemical vapor deposition. In our case, using reactive co­
evaporation, the deposition is over in about 4 min. for a 2000A thick film,
resulting in a deposition rate of 500A/min or about 8A/s, which is very fast.
Using the slower sputtering technique
,
Burmann et al. [8] investigated the
influence of the deposition rate on the orientation, used deposition rates
between 0.2A/s (12A /min) and 1.8A/s (108A/min), or 20 min to 3 hours for a
2000A thick film. We usually follow the deposition by an anneal lasting for 30
minutes at a pressure of 15 torr.
Because of these differences in preparation methods, one can consider
the crystals to be in equilibrium and having had enough time to equilibrate,
while the films might not have had time to equilibrate. This is also obvious
in the published papers, where the papers on crystals are more likely to agree
with each other, while papers on thin films are very dependent on the
preparation method used and the treatment after deposition. In addition,
published papers on films and crystals are not necessarily in agreement with
each other, showing that films are more disordered than crystals and results
from films should be considered within the context of the preparation
method.
1.3 Examples of Differences between Films and Crystals
The oxygen content of YBa2Cu306+ is very important, because it is
strongly related to the properties of YBa2Cu306,. There are several established
methods to measure the oxygen content in crystals, some of which are
available only for crystals. The most common ones for crystals are idiometric
titration and graviometry. For idiometric titration, see Refs. [9] and [10], the
9
sample is dissolved in an acidic solution of KI. This results in the reduction of
all copper to solid Cu(+I) iodide. Back-titration of the liberated iodine by
thiosulfate gives the average oxidation state of copper, from which the oxygen
content of the sample can be calculated. In order to determine the oxygen
content by graviometry, one measures the weight loss while the sample is
heated in an nitrogen atmosphere (this is done especially when the
exploration of the phase-diagram is of interest) or in a reducing atmosphere.
In that case, the reduction products have to be known [9]. Neither method
works for films, because a thin film does not provide enough material. Either
method can be used to calibrate other methods which can be used on films.
Several groups doing research on oxygendeficient films usually had some
bulk material in the vicinity when they were taking oxygen out of films, and
used the bulk material to establish the oxygen content. Apart from providing
a means to establish the oxygen content, the bulk material stabilizes the films
and prevents decomposition [11].
10
1.3.1 Tc versus Oxygen Content
100
crystal
80
film
$
60
T[K]
40
20
0
B
6.2
t
41,
6.3
1
II
I
1
6.4
II
ABIllt
I
6.5
6.6
6.7
oxygen content
I
It
I
II
6.8
I
6.9
7
Fig. 1.4 Tc vs oxygen content for crystals and films. The crystal curve is from
Ref. [5] and is taken from a fit to their measured data; the oxygen content
determined from weight loss. The film data is from [12] who infer the oxygen
content from the anneal procedure and are careful to refer to it as "nominal".
Tc is greatly influenced by the oxygen content. Usually the Tc versus 6
(oxygen deficiency) 2 curve shows two marked plateaux, one at 90 K, for 6 <0.2,
the other at 60 K, for 0.3 <6<0.5, with the material becoming semiconducting
for 6 >0.65 [5]. In films this behavior seems to be less pronounced and there
are papers published which show no plateaux at all. A good example is in Ref.
[12]. Fig. 1.4 is a comparison between the data published by Jorgensen et al. [5]
and Osquiguil et al. [12]. For the films, the plot of Tc versus 6 looks linear and
there is no hint of a plateau. Taking the same T one would determine two
2 6 is normally used to indicate the oxygen deficiency from a fully oxygenated
07; x is used to indicate the oxygen content, from the oxygen-deficient 06.
Occasionally, x is also used to indicate the total amount of oxygen.
11
very different oxygen contents for a crystal or film sample. Osquiguil et al. [12]
are very careful about calling their oxygen content nominal and state that the
same behavior has also been observed in pellets of YBa2Cu306,, which were
quenched from temperature between 650 °C and 550 °C, depending on the
annealing atmosphere, either pure 02, air, or 5%02, 95% N2 [13].
1.3.2 c-Axis Length versus Oxygen Content
The c-axis length is a very reliable indicator of the oxygen content in a
crystal, yet in films it should be used very carefully. A good example is from
Ref. [14], which compares the c-axis length of oxygendeficient films with the
same length for oxygendeficient crystals given in
Ref.
[5]3.
In the
comparative graph, the c-axis length for films is greater by about 0.03 to 0.04A
for the whole range of oxygen contents.
The usual suspects for the wellknown expanded c-axis length of
YBa2Cu306,. thin films are strain, because the film is mismatched to the
substrate, and cation disorder, though oxygen disorder might also be possible.
Ref. [15] has an indepth discussion about the possible reasons for the c-axis
length expansion, summing up the general reasons. They consider
insufficient oxygen unlikely because of high Tc's (= 90 K) and exclude substrate
strain because the a- and b-parameters are not expanded. They claim that
extended defects are unlikely to cause the expansion, because that would give
a distribution of lattice parameters. This leaves them with cation disorder
3 As an interesting aside, the c-axis length expands if oxygen is taken out. This
seems to be counterintuitive, but has to do with the fact that the Cu-0 planes,
which are buckled in the fully oxygenated material, straighten with
decreasing oxygen content[5].
12
(especially Ba substituting on the Y site) as the likeliest candidate for the c-axis
parameter expansion.
Another paper that nicely shows how much one has to be careful with
determining the oxygen content of films is Ref.
[16].
Using Raman
spectroscopy, they find in several of their films the oxygen content to be > 7 if
they use the published empirical formula (for example in Ref. [17]), which
determines the oxygen content from the frequency of the 0(4) line, the 500
cm"1 line. From their c-axis length (which they seem to compare to crystal
data), the film oxygen content is below 7, which makes more sense. They also
attribute these discrepancies to strain and disorder.
1.4 Role of Oxygen
As is obvious from the preceding paragraph, oxygen is very important
in YBa2Cu306,x and a lot of research has gone into investigating oxygen. The
oxygen content is tied to electrical and structural properties and especially the
electrical properties are important from an applications point of view. In
addition, because the oxygen content can vary between 6 and 7, the question
arises about how the oxygen atoms are arranged. It is also noteworthy that the
structure actually provides space for 9 oxygen atoms, evident in some of the
earliest papers, where the formula YBa2Cu309_8 was used, and even if one
neglects the possible oxygen atoms on the Y level, one could naively expect
the structure to have 8 oxygen atoms in it. Ref. [6] gives a short sketch on how
the scientific community realized that YBa2Cu307_8 was the correct formula
instead of YBa2 CU 309-8'
13
1.4.1 Chain Length and Charge Transfer
In the completely oxygenated structure the planes are the place where
the electronic transport takes place, but the chains play an important role in it.
The hole concentration in the planes depends on the oxygen concentration
and ordering in the chains [18]. Transfer of charge from the chains to the
planes increases the hole concentration in the planes [19]. The chains are
therefore considered a "charge reservoir". In order to do that the chains need
to have a certain length. Ref. [20] shows that a minimum chain length of four
Cu-0 segments is required in order to initiate the charge transfer from the
chains to the planes.
1.4.2 Oxygen Ordering: Experiment and Theory
The fully oxygenated YBa2Cu3O7 has only half of the possible oxygen
sites in the basal plane filled and the more oxygen is taken out, the more
possible sites are available to the remaining oxygen atoms. These oxygen
atoms can be arranged in ordered ways and the arrangement of the oxygen
atoms is of great importance, but experimental and the theoretical research
have been difficult to reconcile. Oxygen is very hard to image directly and
experimental techniques usually rely on diffraction techniques which need
order over rather large areas (several 100 A). Computational techniques, on
the other hand, consider the atomic level and, if anything, might have a
problem of covering too small an area due to limited computing power.
There are two main models used to explain oxygen ordering which
yield two different kinds of ordering. Before explaining the models in more
14
detail, one has to clarify the interactions between the oxygen and copper
atoms. Fig. 1.5 shows the interactions used by the models.
V, is the potential between two oxygen atoms at 90° to a copper atom
(nearest neighbor interaction). V3 is the potential between two oxygen atoms
across the unit cell, and V2 is the potential between two oxygen atoms
mediated by a copper atoms; both are next nearest neighbor interactions. V,
and V3 are assumed to be repulsive, but V2 is the potential which is
responsible for the difference between the different models and can be either
attractive or repulsive.
In
the
ASYNNNI
model
(asymmetric-next-nearest-neighbor­
interaction Ising model), first proposed in Ref. [21] the potential V2 is assumed
to be attractive, with the conditions V2 <0<V3<V1. As a consequence, this
model predicts that only long chains are stable and for example predicts a
doubling of the unit cell for x=0.5 with every second chain unoccupied with
respect to the fully oxygenated material (which follows from the weakly
repulsive V3).
V2
Fig. 1.5 Definition of the potentials between oxygen and copper atoms. Filled
circles represent Cu atoms, white circles oxygen atoms.
15
In another widely used model proposed in Ref. [22], there is a strong
repulsion between all the oxygen atoms, including V2. As a consequence, this
model leads to the formation of only the basic 0Cu-0 fragments, but not the
formation of any kind of chains.
The difference between the two models can be seen most clearly if one
looks at the prediction of the order for an oxygen content of 6.5. In the case of
the ASYNNNI model the oxygen atoms order in long chains, with every
second chain missing with respect to the fully oxygenated structure. For the
second model, the oxygen atoms order in the so-called herring-bone structure
which is distinctively different from the chain structure, as can be seen in Fig.
1.6.
It seems that it should be fairly simple to decide between these two
models, especially around an oxygen content of 6.5, because the ordering
0Cu-0 fragments: x = 0.5
ASYNNNI: x = 0.5
O
O
0
0
0
.0.
0
0 00
0
0 4) r_,N
.
._.
0
O
O
0
0
0
(PO
o
.0.
(Th.
...1
0
0
0
(:).
0.
Fig. 1.6 Examples of the ordering of oxygen atoms in the basal plane for an
oxygen content of 6.5. Filled circles represent Cu atoms, the open circles
represent oxygen atoms. The ASYNNNI model predicts long chains, with
every other row unoccupied. For 0Cu--0 fragments, the predicted structure
is reminiscent of a herring-bone weave.
16
proposed by the ASYNNNI model clearly has an orthorhombic structure to it,
while
the
other
model
has
an
average tetragonal
structure.
Yet
experimentalists have had difficulty establishing the validity of either of the
models. This is mostly due to the fact that these techniques have to rely on
diffraction and therefore average over larger areas, while in reality, the order
can be fairly easily disturbed, for example by grain boundary and dislocations.
Some experiments indicate that there also might be a cross-over in the
interaction V2 from being repulsive to attractive, depending on the size of the
rare earth ion [20].
An additional ordering phenomenon in YBa2Cu306, is the so-called
room-temperature annealing. Even at rather low temperatures of roomtemperature and slightly above, the oxygen atoms can rearrange. This in turn
affects the transition temperature and X-ray spectrum [23] and happens for the
orthorhombic as well as the tetragonal structure. Ref. [23] shows the example
of a YBa2Cu306.45 sample that took several days to reach its final T, of = 44 K
(starting at 35 K) at room temperature and of another YBa2Cu30645 sample
that took several weeks at 0 °C to reach its final T, of = 46 K (starting at 39 K).
Furthermore, in x-ray diffraction studies, they see a sharpening of the (400)
and (040) peaks over time for samples with different oxygen contents. This is
most pronounced for the samples with an oxygen content of 6.30 (tetragonal)
and 6.35 (orthorhombic) and happens over = 40 h. In neutron diffraction
experiments ([24]), a relaxation of the a- and c-axis parameters is seen, again
indicative of oxygen ordering.
17
Chapter 2
Sample Preparation
2.1 Thin Film Deposition Techniques
Shortly after the discovery of the HTSC's, scientists started to grow thin
films of the material. Several different techniques have been used, each with
its own set of advantages and disadvantages. One usually differentiates
between physical (sputtering, MBE, e-beam) and chemical (MOCVD)
deposition techniques. During a physical deposition, generally no chemical
reactions occur (but there are some methods during which a chemical
reaction occurs, especially if oxygen is introduced). During a chemical
deposition, there always is a chemical reaction. In addition, the deposition can
either happen in the vapor phase or in the liquid phase. In this discussion, I
divide the different techniques as applied to YBa2Cu306, into two groups
according to the control of stoichiometry.
The first group contains techniques which use a target made of
YBa2Cu307_8 in the right composition, which is evaporated by some means.
For example, ions may bombard the target and therefore transfer momentum
to the surface atoms. This is known as sputtering. In the case of laser ablation,
a laser is directed at the target and evaporates the material. The plume created
then deposits the YBa2Cu306, on the substrate. The advantage, and at the
same time, disadvantage, of these methods is that the target has to have the
right composition. In some methods, the stoichiometry of the target is
transferred to the film. In others, some target species are preferentially
transferred, so that the film has a different stoichiometry. This difference may
vary with deposition conditions, and has to be carefully established. Ref. [25]
18
shows that using a stoichiometric target (Ba/Y = 2), gives the best films if an
oxygen pressure of 80 mtorr during sputtering. If they decrease the oxygen
pressure to 5 mtorr, the best films are deposited if the target has a Ba/Y ratio
of 1. They do not do any compositional analysis on the film itself.
The second group of deposition techniques evaporates the constituent
materials separately. In the simplest case, the evaporation is achieved by
placing the constituent metals into metal boats and resistively heating them.
The evaporation can also be effected using electron beams or Knudsen
effusion cells, which produce very directed evaporant streams. Because the
metals are evaporated individually, the composition can be changed during
the deposition. On the other hand, three rates have to be kept constant in the
case of YBa2Cu307_5 as opposed to just one in the case of the techniques like
sputtering from one target.
Chemical vapor deposition (CVD) belongs into the second group. CVD
is usually used to provide coatings over large areas. In order to deposit
YBa2Cu3078, MOCVD (metallo-organic CVD) is usually used. Because the
carrier gases are based on carbon derivatives, films deposited by MOCVD
usually contain carbon impurities.
2.2 Reactive Co-Evaporation
Our films are deposited by a technique called reactive coevaporation.
Evaporation of the material can be achieved by several means. In our case, the
metals, Y, Ba and Cu are evaporated from metal boats, depending on the
material either W (for Y), Ta (for Ba) or Mo (for Cu). These boats are
resistively heated and the evaporation rates are kept constant with the help of
feed-back loops. An important point to keep in mind for the design of an
19
evaporation system is that the evaporant does not uniformly coat the
substrate, but the thickness falls off from the center according to the following
formula (for a small area source):
d0
2 -2
.[ ±(1)
_
h
(2.1)
with d the thickness at a point I from the center of the substrate and h the
substrate-source distance [26].
"Coevaporation" refers to the simultaneous evaporation of materials.
This adds problems mostly from a design point of view. The advantage is that
film stoichiometry can be easily varied, but it adds design complexity. Several
sources have to be regulated at the same time. This either means using a
technique which can pick the different species out of the evaporant stream
(like mass-spectrometry) or assuring that each detector is exposed to the
stream from only one source (our technique).
"Reactive" refers to the introduction of oxygen close to the substrate to
facilitate formation of the oxide. This is unusual; most evaporation
techniques try to use as low a background pressure as possible because the
mean free path decreases with increasing pressure from 500 cm at 10' torr to
5 cm at 10' torr and 5 mm at 10' torr. For YBa2Cu307_8 it was shown very early
on, that the "insitu" process, whereby oxygen was introduced during the
deposition in order to grow the material, produced films with far better
structure and qualities than those produced by "exsitu" processes where the
oxide formation was performed after invacuo deposition. The insitu
process seldom requires local pressures greater than 10-4 torr, where the mean
free path is still of the order of 50 cm. This means that even though collisions
between the evaporated metals and the oxygen might occur in the gas phase
resulting in the formation of metal oxides, the probability is very small.
20
Instead, the formation of the material will take place at the substrate, which is
exposed to a high incidence of metal and oxygen particles [26].
2.3 Factors Influencing the Film Growth
It should be already clear that there are several parameters which
influence the film growth and therefore the quality of the film grown. The
most important factors are the stoichiometry, evaporation rates, the
temperature of the substrate during the deposition, the oxygen pressure
during the deposition, the substrate temperature during the annealing and
pressure during the annealing. Several additional factors can influence the
quality of the film, but are not well controlled in our system, like the quality
of the substrate, and the surface cleanliness. In addition, these parameters are
not all independent. The following gives examples of what the individual
parameters mostly influence.
2.3.1 Stoichiometry
Even though it seems obvious that the individual rates should be set as
to achieve the perfect 1:2:3 stoichiometry, Dr. Dennis Tom in his thesis [27]
showed that the best film are films which are slightly copper-rich.
2.3.2 Substrate Temperature During the Deposition
This is the crucial factor in order to achieve the desired orientation.
The substrate temperature has to be higher to grow c-axis oriented films and
lower to grow a-axis oriented films (the actual temperatures strongly depend
21
on the technique used, but there
is =100 °C difference
for the two
orientations). On the other hand, it has to be high enough (the lower limit
was never actually determined, but = 550 °C is probably a good guess) to grow
films that are oriented.
2.3.3 Oxygen Pressure During Deposition
Some oxygen (pressure depends on the technique used) has to be
supplied during the deposition, otherwise the material will not grow.
Depositing just barium, copper and yttrium at the same time without oxygen
and then annealing it in oxygen will grow an amorphous material (as
evidenced unintentionally by a PAC run) rather than oriented YBa2Cu306.,
2.3.4 Postanneal: Oxygen Pressure
This parameter influences the length of the c-axis parameter and it also
changes the oxygen content. There is a known correlation between the c-axis
length and the oxygen content for crystals, but because films are more
disordered than crystals, this relationship cannot be simply carried over, but
has to be carefully checked, as illustrated in Chapter 1.
2.3.5 Postanneal: Substrate Temperature
Not much investigation was done by me into this parameter. In the
case of the tetragonal YBa2Cu306,25 films, which were produced by adding an
anneal in vacuum after the "standard" deposition, the higher the substrate
22
temperature is during the anneal, the more we produce of the impurities
BaCu2O2 and CuYO2.
2.4 Description of the Evaporation System
A very detailed description of the evaporation system is given in Dr. D.
Tom's thesis [27]; I will only give a brief overview in this chapter. A more
detailed description, describing the relevance and possible choices at each step
of the deposition process, can be found in Appendix A.
The film deposition is accomplished by evaporating the constituent
metals individually from metal boats which are resistively heated. In order to
control the rates, a quartz crystal is mounted above each boat and shielded
from the evaporant stream of the other boat. (This is important, because the
quartz crystals can not distinguish between the different materials. Their
operating principle is to detect a change in frequency which relates to a change
in weight). The crystals are connected to Sycon4 quartz rate monitors, which
regulate the current supplied to the boats via a feedback loop. For PAC
experiments, a fourth source is added, which also is resistively heated, but is
shaped more like a gun or cannon in order to supply a directed stream of 111In
atoms at the substrate. The substrate is mounted in a ringshaped annulus,
which supplies the oxygen locally at the substrate. The amount of oxygen
supplied during the deposition is regulated by a mass flow controller. A PBN
heater is placed on the oxygen annulus, heating the substrate to between
550 °C and 650 °C and the annulus to a somewhat lower temperature. Because
STC-200 Deposition Rate Controller: Sycon Instruments, Inc., 6757 Kinne
Street, Syracuse, New York, 13057
23
the rates take some time to stabilize, a shutter is mounted between the
substrate and the boats and is opened only after the rates are stable.
2.5 Control of Oxygen Content
In order to reliably control the oxygen content of YBa2Cu307_8 films, an
apparatus was built, drawing mostly on ideas provided in Refs. [11] and [12].
The basic idea in both papers is to enclose the film in a quartz tube with bulk
material and keep it at certain temperatures and pressures according to the
phase diagram established by Ref. [28] and reproduced in Fig. 2.1. The lines
indicate the temperature and pressure required to maintain a particular
values of 6+x and are later referred to as phase lines. The bulk material is
used as an oxygen reservoir and helps to establish the desired value. In
1.5
1.0
0
6.4
6.\
6.5 6.6
6.7
6.8
6.9
0.5
0
-5= -0.5
1.
cs -1.5
-J
-2.0
-2.5
-3.0
-3.5
0.8
0.9
1.0
1.1
1.2
1.3
1.4
1.5
1.6
1000/K
Fig. 2.1 Phase diagram used to change the oxygen content of our films. It was
proposed in Ref. [28] for bulk.
24
addition, Ref. [11] found that if they did not have bulk material in the vicinity
when they tried to achieve high oxygen deficiencies, their films decomposed
completely.
Both methods start out by evacuating the quartz tube to 10-6 Torr.
Osquiguil then backfills the tube with 10 Torr of oxygen. The anneal
temperature therefore is determined by the pressure in the quartz tube and
the desired oxygen content. After the system has equilibrated (the pressure
stays constant), the sample is cooled down following the appropriate phase
line, and eventually quenched.
In the method used by Ref. [11] the sample is heated up to 400 °C, where
a rapid rise in pressure is observed when the oxygen becomes mobile. Then
the valve connecting the quartz tube to the pump is closed and the sample is
heated to 600 °C, where the pressure is adjusted to the value appropriate to
the desired oxygen content.
2.5.1 Description of Apparatus
The main component of the system I developed in this laboratory is an 11"­
long, I/2"-diameter quartz tube, connected via a glass-to-metal seal to a
stainless steel KF-flange, which in turn is connected to five-way cross. The
length of this quartz tube was determined by an existing tube furnace used to
heat the sample. The "sweet spot" of the mainly furnace is about 8" into the
furnace. With the length chosen for the quartz tube, the glass-to-metal seal is
then far enough from the furnace that the seal's integrity is not breached'
5 This was shown inadvertently when the thermocouple shortened out right
next to the glass-to-metal seal and therefore was reading the temperature
there. Even though the furnace was red hot (>600 °C), the thermocouple was
reading not more than 30 °C.
25
(quartz is not a good conductor of heat). Besides the quartz tube, a Pirani
gauge (reading from atmosphere to 104 Torr), a K-type thermocouple
feedthrough, a needle valve connected to an oxygen tank, and a Varian turbo-
pumping station are connected to the cross, all with NW 40 KF-flanges. This
set-up facilitates the easy loading of the films.
temperature
controller
417
thermocouple
feedthrough
-.A /\
to 0, supply
to pump
double valve
set-up
Eumace
pressure
gauge
YBa2Cu30,., film on substrate
YBa2Cu30, pellet
Fig. 2.2 System used to control the oxygen content
26
Between the pumping station and the cross are two block valves which
serve the same purpose as a throttle valve by having both of them barely
opened when trying to stay on a given phase line. The twovalve system
allows finer control of the oxygen pressure, and with both of the valves barely
opened, the one closest to the pumping station is used for the coarse
regulation, while the one closest to the sample does the fine-tuning. With the
twovalve setup, a wide range of pressure changes can be regulated, from
e.g. 18 torr/min needed during part of the anneal for an oxygen content of 6.7
to 0.5 mtorr/min required during part of the anneal for an oxygen content of
6.4. The cooling-rate is about 6 °C/min, determined by how fast the furnaces
cools down by itself. The temperature is controlled via a type K thermocouple
and a Eurotherm 808 temperature controller6, with the tip
of the
thermocouple in close proximity to the film. With the current PID (P = 25, I =
12, D = 0) setting, the furnace has a stability of about -2 °C and +5 °C. The cross
and quartz tube are mounted on a cart in order to facilitate quenching the
film by pulling the set-up out of the furnace. Fig. 2.2 is a sketch of the system
described above.
2.5.2 General Description of the Procedure
The steps to deoxygenate a film are the same, regardless of the desired
final oxygen content. First a film and a pellet of YBa2Cu307, pressed in Dr.
Sleight's lab from powder bought from Alfa Aesar7 are put into the quartz
tube. The tube is connected to the rest of the apparatus, making sure that the
6 Eurotherm Corporation, 11485 Sunset Hills Road, Reston, Virginia 22090­
5286
YBa2Cu3O powder, stock # 39534, lot # K16F14 from Alfa Aesar, 30 Bond
Street, Ward Hill, Ma 01835-8099
27
thermocouple is right next to the sample and does not short electrically along
its distance. Then the whole system is evacuated to <10 torr and heated to
600 °C. The controller is set to ramp at 10 °C/min, but for temperatures lower
than 250 °C, the furnace has a tendency to overshoot. During the heating, one
usually (not always, if a pellet is reused) observes a rise in pressure at
temperatures slightly less than 100 °C, when adsorbed water vapors start to
evaporate. The pressure does not go any higher than 10-2 torr and goes back
down to below the detection limit (10' torr) of the Pirani gauge. Because the
furnace does not regulate very well for these "low" temperatures, the heating
is continued. The system continues to warm, while being evacuated, till about
400 °C, when there is another rise in pressure, which can be explained by the
oxygen becoming mobile and leaving the film and pellet. At this time the
valve closest to the tube is quickly closed. The pressure in the system rises as
the film is heated to about 600 °C (for PAC films, 550 °C was often used), as
more oxygen is released from the film and pellet. After reaching 600 °C,
depending on the oxygen content desired, the system either has to be
backfilled carefully with oxygen or the pressure has to be equally carefully
reduced. This is done by opening both valves just barely. If the system has to
be backfilled, the lines are flushed with oxygen beforehand in order to ensure
that only oxygen is being admitted to the system. The pressure is adjusted to
the value determined by the temperature (600 °C) and the desired oxygen
content. The anneal at 600 °C lasts for about 1 hour, after which the
temperature and pressure are adjusted to the appropriate phase line, until the
temperature is 450 °C, where it is held constant for 1 hour. The final step is
cooling the system down to 400 °C, again staying on the phase line, and then
quenching by quickly pulling the whole apparatus out of the furnace.
28
The temperature of 600 °C for the initial anneal step was chosen
because it is high enough to ensure adequate oxygen mobility to reach the
equilibrium, but not high enough to cause relaxation of crystallographic
defects [11]. Except for film HTC 32, the higher anneal temperature for the
PAC films was 550 °C, because there was evidence of a impurity site (also
referred to as B site) forming in the spectra.
Quenching from 600 °C does not ensure that the film will have the
desired oxygen content, because the pressure does not drop fast enough and
the film might incorporate some additional oxygen. The additional step at
450 °C is therefore added to ensure that the film has the desired oxygen
content. At this temperature the oxygen mobility is still high enough to
ensure that the film will be in equilibrium.
The oxygen contents that can be easily achieved with this set-up and
this procedure are between 6.3 and 6.7. Below 6.3 the pressures required for
the 450 °C anneal are below the range of the Pirani gauge, and above 6.7 the
600 °C anneal step would require pressurizing the system above atmospheric
pressure, a task it is not designed for. Anneals with a lower initial
temperature have not been tried yet, but they should make it possible to
obtain films with oxygen contents between 6.7 and 7.0.
29
Chapter 3
Characterization
3.1 Raman Spectroscopy
Raman spectroscopy is one of the techniques used to quantify the
oxygen content of YBa2Cu306, and determine the orientation of YBa2Cu306,
thin films and crystals. The basic principle behind Raman spectroscopy is the
inelastic
scattering of photons off phonons from the structure under
investigation. The energy difference between the incoming and scattered
photons excites the lattice vibrations. Because there are so many atoms in the
YBa2Cu306, unit cell, there are several possible excitations, the main ones
used are listed in Table 3.1.
Spectra are commonly labeled by the Porto notation z(x,x).2, which
describes the polarization geometry. The first and last letter describe the
propagation direction of the incoming and scattered photon, the letters in
frequency [cm-1]
atoms responsible
118
Ba-Ba motion
145
Cu-Cu (in plane) motion
335
0(2)-0(3), outofphase vibration, along the c-axis
440
0(2)-0(3), in-phase vibration, along the c-axis
500
0(4), vibrations along the c-axis
Table 3.1 Raman frequencies (in cm 1) for YBa2Cu3O7. Frequencies are taken
from Ref. [29].
30
brackets indicate the incoming and scattered polarization of the electric vector
E [30]. x is parallel to the a-axis, y is parallel to the b-axis and z is parallel to the
c-axis. Sometimes x' and y'axis are used, which are oriented at 45° to the x
and yaxis.
The line at 500 cm -1 is the most sensitive to oxygen content and is most
commonly used to determine the oxygen content, based on an empirical
formula [31], using the frequency co(x) (in cm-1).
x = 0.037(o(x)
11.555
(3.1)
This line shifts approximately linearly down from 500 cm -1 to 475 cm-1 with
decreasing oxygen content [29]. Thomsen et al. [29] also quote results from
Macfarlane et al. [32], which differ somewhat from their own. They are in
agreement about the 500 cm -1. line and the 145 cm-1 line, but for the 335 cm -1
line, Thomsen finds a linear change by about 10 cm 1, while Macfarlane finds
no change at all. For the 440 cm-1 line, Thomsen finds the line to change
linearly with oxygen content, while Macfarlane finds that the frequency
decreases with decreasing, oxygen content till about 6.5 and then stays
constant. They attribute the differences to sample preparations and the nearsurface oxygen concentration.
As an example, Fig. 3.1 shows Raman spectra taken on c-axis oriented
films deposited on SrTiO3 substrates. These spectra were taken by Dr.
Christian Thomsen's research group at the Technische Universitat in Berlin
and are of an orthorhombic and two tetragonal films from our laboratory.
Features in the spectra below 340 cm -1 are an artifact of the filters used. The
627 cm-1 is strong because the excitation was done with red light and the
645 cm -1 and 417 cm-1 are consequences of this excitation and are a Froehlich
induced resonance.
31
Because polarized light is used and only certain polarizations excite
certain vibrations, this method can also be used to determine the orientation
of the film. Thomsen et al. [33] developed a method to determine the epitaxy
of c-axis oriented films, Dieckmann et al. [30] expanded upon it and used it for
both a-axis and c-axis oriented films.
Gibson et al. [34] and Cohen et al. [35] showed that the 500 cm -1 line as
well as the 115 cm -1 line and an additional line at 585 cm -1 yield some
information about cation disorder in this films. The frequency of the 115 cm-1
decreases with increasing disorder and the intensity of the 585 cm-1 line
(normalized to the 340 cm-') is larger for higher disorder. They also found that
with increasing disorder the frequency of the 500 cm-1 line decreases, which
can lead to too low an estimate for the oxygen content. On the other hand,
Martinez et al. [16] found a shift to higher frequencies for some of films and
attributed it to departures from the cation stoichiometry. This again shows
that in thin films, several factors contribute to changes in a certain quantity
and it might not always be possible to determine the particular factors in any
given case.
Furthermore, Raman spectroscopy can be used to find impurities, at
least within the penetration depth of the light used. BaCuO2, Y2BaCu05,
Y2Cu2O5, Y203, BaO and CuO are the most commonly found impurities, and
Ref. [29] gives a list of references for the Raman spectra of the impurities.
32
YBa2Cu3
/ SrTiO3
T=300K
X=632.8nm
LABRAM
ortho
tetra (lh
500 t)
tetra (5h at 500 C.)
200
400
600
800
Raman Shift (cm-1)
Fig. 3.1 Raman spectra taken on an orthorhombic film and two tetragonal
films subjected to different anneals. The differences between the
orthorhombic and tetragonal films are very obvious, but the differences
between the two different anneals for the tetragonal films do not show up in
the Raman spectra. The orthorhombic film was determined to be fully
oygenated, and the tetragonal films have an oxygen content of 6.25 (±0.15).
The data was taken by Prof. Thomsen's research group at the Technische
Universitat in Berlin.
33
3.2 Rutherford Backscattering
Rutherford's famous expression for the elastic scattering of charged
particles is the basis of the technique of Rutherford backscattering (RBS)
which is used to determine the composition of materials. It can also give
information about thickness, crystallinity, and homogeneity of films under
certain conditions. A brief overview follows, that extracts the most important
points from [36].
In a nutshell, RBS is the shooting of a charged projectile of known
energy at a target and the measurement of its energy after the scattering.
Because the energies chosen are small enough that nuclear forces are not
important, the interaction is considered entirely Coulombic.
detector
low
depth
Fig. 3.2 Typical Rutherford Backscattering setup.
34
The kinematic factor K is defined as the ratio of the energy of the
projectile after the scattering (Ep) to the energy before the scattering E0 and
using energy and momentum conservation:
^2
41- [(Mr/MT)sine]2 +(mp/mr)cose
K = -2­
(3.2)
Eo
1±(MpliVIT)
It depends only on the ratio of the masses of the projectile (Mr) and the target
(MT) and the scattering angle.
From an experimenter's point of view, the more energy that is lost by
the projectile during the scattering the better, because that makes it easier to
measure and identify the projectile after the scattering. For a given E0, this
means that K should be as small as possible. Leaving the masses fixed, K is the
smallest for 0 = 180°, hence the term "backscattering". In reality, the detector
cannot be placed at 180°, because that would put it into the path of the
incident beam (see also Fig. 3.2), but it is placed at least 170°, or a detector in
the shape of a ring is used.
If two nuclei scatter from each other by the Coulomb interaction (and
Mp <<MT) the differential cross section is:
do'
ZrZTe2 y
dfl
167teoE
1
(3.3)
sin4(0/2)
The scattering cross section depends strongly on the atomic number Z of the
material under investigation and hence serves as a means to differentiate
between the different elements that comprise a material.
If a particle scatters in a depth x, with al and a2 the angles of the
incident and scattered particle beams with the surface, the energy loss is
=
[ K dE
sin al dx
+
E,
dE
sin a2 dx KEG
1
x
(3.4)
35
which means that from the energy attenuation dE/dx in the material one can
calculate the thickness of the film. There is a linear relationship between the
energy loss and the thickness. Points in an energy range A corresponds to
scattering events from a layer of thickness Ax. Fig. 3.3 exemplifies this
relationship. RBS data usually
is
taken by counting the number of
backscattered particles at an energy EP(i) and within an energy interval A
(given by the energy resolution of the detector) and sorting it in different bins
with i being the channel index. In the case of perpendicular incidence of the
projectile (a1 = 90°) and perfect back-scattering (8 = a1 +a2 = 180°), the above
equation reduces to
dE.
= [KdE +
dx E0 dx
d
(3.5)
KE0
Therefore, by measuring AE, one can measure the thickness of the film or
layer under question.
KE o
Fig. 3.3 Sample histogram of a thin self-supporting film
36
Because the kinematic factor K depends on the mass of the target, if a
material contains several different kind of atoms, the leading edges will be at
different energies, which makes the identification of different materials
possible, unless the atomic numbers are too close for the energy resolution of
the detector. The thinner the material, the better separated are the peaks from
different kind of atoms.
Additional information can be gained by integrating the area under the
curve. Starting from
H(i) =
dI
A.Q to =
/ N. AS2t
do-
(3.6)
E(x1) A
one can calculate the area F under the curve and find the following equation:
do-
F = /000tonddi2
(3.7)
E0
with 4the beam intensity, to the measuring time, AS2 the detector solid angle
and n the target particle density. The thickness can therefore be calculated if
all the terms in the equation are known.
Additional information about the stoichiometry can also be gained
from the above equation. One can get absolute stoichiometry (at least within a
factor of d) because all the terms in equation (3.7) except nd are known. By
forming the ratio for two kinds of atoms, most of the factors cancel and one is
left with terms relevant to the different kind of atoms, e. g. taking atom 1 and
atom 2
F
F2
n ( Z1
n2 \Z2
2
(3.8)
This formula gives only ratios, and in order to get absolute number, the
additional condition for YBa2Cu306, is imposed that the cations have to add
up to 6.
37
Our RBS analyses are performed at the University of Arizona by
Professor L. C. McIntyre. The aparticles used are accelerated to 2
4 MeV,
and the detector is positioned at 0 = 170°. With this set-up, an accuracy of 1%
on the relative stoichiometry and 3% on the absolute stoichiometry can be
achieved
if the peaks are separated. This requires a thin film
2000A for
YBa2Cu307_5) on a lowZ substrate like MgO. RBS showed that our best films
tend to be copperrich.
In Rutherford backscattering, a phenomenon known as channeling
can be used to gain structural information. Due to the periodicity of a lattice, a
charged particle can experience many small glancing collisions and be guided
deep into the crystal along atomic rows or atomic planes (channels). The
better the crystallinity of the material under investigation, the deeper the
charged particles will penetrate the sample and the fewer will be detected in a
detector set up for Rutherford backscattering. xmin, the ratio between the yield
for a random spectrum and an aligned spectrum, for our films are about 20
50 %, which is large compared to 3 8 % for very good films. Ito et al. [37] used
channeling to investigate a-axis films of different thickness and found that
the xnur, value increased with increasing film thickness, indicating strain
relaxation along the c-axis, which, for a-axis films, is in the substrate plane.
Another application of channeling is probing the orientation of our films
relative to the substrates and it has been found that the MgO substrate normal
and the film normal are misaligned by
1°.
3.3 The Transition Temperature
In order to measure T the critical temperature at which the resistance
of a superconductor drops below the noise level, a four-probe measurement is
38
used. The design of the probe has been described extensively in Dr. D. Tom's
thesis; I will give just a quick summary.
The sample is mounted on a copper block with the help of four Be-Cu
wires, which are bent to act as springs. These wires keep the sample in place
and also provide the electrical contact with the outer two supplying a current,
and the inner two measuring the voltage. This 4terminal measurement
eliminates lead and contact resistance. The copper block houses a Pt resistance
thermometer. The temperature is measured by monitoring its resistance,
which decreases linearly with decreasing temperature over the range 30 K to
about 500 K. With the help of the calibration curve, the resistance is translated
into a temperature. The range of a Pt thermometer can be extended up to
800 K and down to about 14 K; for temperatures lower than that it becomes
insensitive to temperature changes. The copper block is used to ensure that
the thermometer is at the same temperature as the sample and that the
sample temperature is uniform'.
This whole unit is mounted in 1inch stainless steel tube and covered
with a copper shield. The shield is used to protect the sample from refrigerant
splashing. With different adapters, the probe can either be mounted in a
dewar and used with liquid nitrogen, or it can be used on a transport dewar
filled with liquid helium. In both cases, the temperature is changed by
making use of the temperature gradient between the outside and the boiling
refrigerant.
The measurement is controlled by a computer program written in
LabVIEWTM 9. The sample voltage is measured twice, first with the current in
'A good example for that is seen during the warm-up of the probe after a
measurement has been taken. As soon as the probe is exposed to air, it frosts
over. During the warm-up, the frost on the copper-block melts all at once,
very differently from the frost melting on the rest of the probe.
LabVIEWTM is a trademark of National Instruments Corporation.
39
one direction, then in the other. The two voltages are subtracted to eliminate
thermal voltages10 and then divided by two. Then the resistance is calculated
according to R=U/I. At the same time, the temperature is measured twice and
averaged, to compensate for any cooling of the sample during the two voltage
measurements.
A good measurement yields a very smooth curve with instrument
limited noise, which means that for a good c-axis oriented, orthorhombic
YBa2Cu3O7 film with decreasing temperature the resistance has to decrease for
each data point. Noisy data are an indication of some form of trouble, usually
not with the sample, but rather with the contacts as can be seen in Fig. 3.4,
which shows data for a film, where the contacts became faulty around 90 K.
Even the data for higher temperatures is not as smooth as it should be for a
good Tc measurement. Measuring the resistance across any of two pairs of
contacts with the sample in place should give about the same resistance for all
the different pairs. If one contact gives a significantly higher resistance than
the others (by comparing the different pairs of wires), one can try to carefully
bend the wire to give it more tension, but very often one has to resolder the
wire.
The high resistance is an indication that the wire is barely making
contact with the sample surface, and that it is very susceptible to mechanical
disturbances, like moving the sample further into the dewar. This shows up
as large changes in the resistance and random readings if the contact is lost
altogether. Sometimes these measurements can be used to get a rough feel for
where the transition temperature is, but they cannot be used for much more.
1° From the first measurement, one gets Usampie + Uthermabthe second
measurement yields Usampie Uthermal, because the thermal voltage does not
change with the reversal in current. Taking the difference Usampie Utherrnal(
Usample
UthermaD= 2Usample
40
Much information can be gained from the resistance measurements.
Most important are the transition temperature, where the resistance drops
below the noise level, and the width of the transition, which gives an
indication of homogeneity. Information about the nature of the carriers and
the normal state can be obtained from the general shape of the transition
curve, but we never explored that venue.
For a good orthorhombic film, the transition should be > 90 K with a
width of
0.5 K. In addition, the general shape should be decreasing linearly
with temperature and extrapolate to zero like a metal, which leads to a ratio
of R(300 K)/R(100 K) = 3. Because most of my research was concerned with
films deliberately deposited under non-ideal conditions, most of the curves I
took did not have this shape. This is especially evident in the case of the aaxis films' R(T) curves, which in most cases are almost flat with decreasing
temperature. They do not have the slope of 3 required for a good metallic film
and in many cases show a slight increase just before going through the
2
"I"
I
1.5
F
11.
0.5
CC
r
0
.
-0.5
1
75
80
90
85
95
100
T(K)
Fig. 3.4 Example of a R(T) curve, where contacts became faulty around 90 K.
41
transition. Fig. 5.6 is an example of a c-axis film (though not optimal) and an
a-axis film which clearly shows the difference in shape and is an example of a
good Tc measurement, compared to the example in Fig. 3.4.
Transition temperature measurements are a strange hybrid, neither a
bulk characterization nor a truly local characterization technique. Because a
continuous, macroscopic path between the contacts is required, it is not a local
technique, but because only a single path out of the multitude possible is
necessary, it is not a true bulk technique like AC susceptibility, where the
whole sample contributes to the signal.
3.4 X-Ray Diffraction
X-ray diffraction is one of the main tools I used to characterize my films
and it is very widely used in thin film science in general. It is used to gain
information about the structure and the orientation of films. It can also be
used to get information about disorder, but x-ray diffraction's forte is the
investigation of order, which means it can tell us only about small deviations
from order.
There are two different descriptions of x-ray diffraction, one associated
with Bragg, the other with von Laue; Ref. [38] show the equivalence of these
two points of view quite nicely. Both descriptions consider elastic x-ray
scattering (the wavelength does not change in the case of the Bragg condition
or I k 1= 1k' I when using the von Laue condition). This scattering is from a
periodic lattice and therefore when the x-rays interfere constructively, a peak
in the scattering intensity is observed. The difference between the two
descriptions is how the source of the scattering is described.
42
In order to derive the Bragg condition, it is helpful to look at a typical
set-up as illustrated in Fig. 3.5. Within the crystal, we can define an imaginary
set of planes, some distance d apart, which diffract the incident x-ray beam.
These planes can be identified by Miller indices, h, k, and 1. The angle of
incidence is assumed to be the same as the angle of reflection. In order to
observe a peak, the x-rays diffracted from all the planes in the crystal have to
interfere constructively, namely the path difference between two successive
planes has to be an integral multiple of the wavelength, which leads to the
Bragg-condition:
nA. = 2d sin 0
(3.9)
It is important to note that even though Fig. 3.5 looks like the incident
beams are diffracting off atoms, the dots are representations of the unit cell
n
X-ray
Source
Detector
-0
Fig. 3.5 X-ray diffraction in 0-26 mode (BraggBrentano geometry). The
source moves through 0, while the detector moves through 20. (More
commonly, the source is fixed and the sample and detector move to maintain
to maintain the 0-20 relationship.
43
and can have many atoms associated with them, as in the case of YBa2Cu306,,
where each dot represents 12 to 13 atoms, depending on the oxygen content.
In the von Laue description, the crystal is regarded as composed of
identical scatterers (sets of ions or atoms), placed at the sites R of a Bravais
lattice. Each can reradiate the incident radiation in all directions. To derive
the von Laue condition for constructive interference of scattered radiation,
we consider Fig. 3.6 which shows the scattering from two points of the
Bravais lattice, separated by d. The condition for constructive interference,
when the path difference is equal to an integral number of wavelengths, is
d01--
(3.10)
Multiplying the above equation by 2n/A, to get from the unit direction
vectors to the wave vectors, and realizing that all d are vectors of the Bravais
lattice, one obtains the following equation
R (k k) = 2nin.
(3.11)
In words, constructive interference will occur if the change in wave vector kk' is a vector of the reciprocal lattice.
n
k'
Fig. 3.6 The path difference between rays scattered from two points of
Bravais lattice is d (ft n )
a
44
The von Laue description is mainly used when dealing with single
crystals, because it provides an easy way to test which peaks can be expected in
a diffraction spectrum, depending on the (range of) wavelength (white x-ray
light or one particular line) used and the orientation of the crystal with
respect to the incident radiation. When working with powder samples, where
all different orientations are present, using the Bragg condition is the easiest
and most intuitive.
To investigate what influences the x-ray spectrum and therefore
provides additional information, it is best to start with an infinite crystal. The
Bragg condition defines where it is possible to expect peaks.
In this case, there is a path difference of an integer multiple of X
between the beams diffracted off the different planes. If the angle is slightly
different from a Bragg angle, there is always a plane somewhere in the
infinite crystal, which has a path difference of exactly X/2 from the plane
under consideration, .and so the diffracted beams from these two planes will
cancel each other. Therefore, in a perfect infinite crystal, one finds infinitely
sharp peaks.
Real crystals are not infinite, and diffraction peaks consequently
broaden, because of the absence of planes to complete the destructive
interference for the off-resonance condition. This is best illustrated by an
example. Imagine a crystal having 1000 diffracting planes. Suppose for an
angle A, the beams diffracted from the 1st and 601" plane cancel each other.
Then the same will happen with #2 and #602, and so on all the way up to
#400 and #1000. Beams diffracted off the planes #401 to #600, on the other
hand, will not destructively interfere and therefore will contribute to the
detected intensity. In an infinite crystal, there still would have been another
45
plane to destructively interfere with. A rigorous treatment results in the
Scherrer formula [39]
B=
0.9A
t COS BB
(3.12)
B gives the breadth (above instrumental broadening) of the curve in 20 and is
defined as the width at half maximum of the peak", t is the thickness of the
sample (or more precisely the thickness of the crystallites) and 9B is the angle
for the Bragg reflection. For a typical film thickness of 2000 A, this leads to a
broadening of the (005) peak at 20 = 38. 42° by 0.042°. The broadening becomes
more important with increasing 013 or decreasing thickness, as shown in Fig.
3.7. This is well below the experimental values of the FWHM of our films.
There are other peak-broadening mechanisms. No x-ray beam is truly
monochromatic; the Ka line has a linewidth of about 0.001A. For X = 1.5A,
and 0 = 45°, this broadens the line by 0.08°. In addition, no real x-ray beam is
truly parallel, which was an assumption in the original derivation. Strain in
the film can cause the peaks to broaden, because it induces variations in the
lattice parameter. Differentiating Bragg's law with respect to 0, one obtains
AO = --Ad tan 0
d
(3.13)
which shows the broadening of the peaks with respect to a variation in lattice
parameter. Since tan 0 approaches infinity as 0 approaches 90°, the influence
on the peak width can be rather marked, which Fig. 3.8 also illustrates.
" Also referred to as FWHM: Full Width Half Maximum
46
0.16
1000 A
0.14
0.12
E.
1500 A
0.1
0.08
2000 A
0.06
2500 A
3000 A
0.04
0.02
40
20
0
60
80
100
120
20 m
Fig. 3.7 Influence of the film thickness on the broadening of diffraction peaks
at different angles. The wavelength of Cu Ka (1.54 A) was used for the
calculation.
1.2
Ac/c=0.1%
Ac/c=0.5%
-Lc/c=1%
film #1
film #2
0.8­
'eV
0.6­
0.4­
0.2-
*
*­
0
20
­
40
60
80
100
120
20[01
Fig. 3.8 Influence of a variation in the lattice parameter on the peak width.
Ac/c=1% is a variation in lattice parameter of =0.1A, Ac/c=0.5% yields =0.06A
and Ac/c=0.1% yields 0.01A for a lattice parameter of 11.7A. Film #1 was
deposited on a MgO substrate with small substrate peaks, film #2 was
deposited on a MgO substrate with strong substrate peaks. The (008) peaks are
missing, because the peaks are too small to fit reliably.
47
Fig. 3.8 also has some data from films deposited in this laboratory. The
two films chosen are "standard" films, except that film #1 was deposited on a
MgO substrate with small substrate peak, possibly indicating an old substrate
that formed an amorphous surface layer, and film #2 was deposited on a MgO
substrate with strong substrate peaks. Fig. 3.8 in conjunction with Fig. 3.7
shows clearly that the main factor responsible for the broadening of the film
peak are variations in the lattice parameter, resulting from strain in the film.
As [40] pointed out, this result is not as undesirable as it might sound, because
there is a 9% lattice mismatch between YBa2Cu306, and MgO and the
adjustment of the lattice parameter is evident in the line broadening and
shows that there is a semicoherent interface between the film and the
substrate.
There is additional information to be obtained from the differences in
intensities of the Bragg reflections, including the case where the intensity is
zero (even though the reflection is allowed according to Bragg's law).
To explain this, we have to put the atoms back into the unit cell. The
phase difference for a wave scattered from an atom at the origin and that
scattered from an atom at the point B with the coordinates (x,y,z) or the
fractional coordinates (u=x/a, v=y/b, and w=z/c) for the hkl reflection can be
written as
0.27r(hu+kv+lw)
(3.14)
Using the complex notation for describing waves, the amplitude of the
scattered wave can be written as
Ael° =
fe2rzi(hu+lcv+Iw)
(3.15)
where f is the atomic structure factor and gives a measure of how efficiently
the atom scatters the incoming beam. It depends on the wavelength of
48
radiation used, the atom, and the angle of detection [41]. The atomic structure
factors are tabulated in Cullity [42].
The resultant wave is obtained by summing the above equation over
all the individual atoms in the unit cell at their respective positions. If a unit
cell contains atoms 1, 2,... N, each having fractional coordinates ul v1 w1, u2 v2
w2, u3 v3 w3,... and individual atomic scattering factors fl, f2, f3,.. then the
structure factor F for the hkl reflection will be:
F
fie2ri(hui +kvi +1w1)
,w2zi(hu2 + kv2 +1w2)
,2iti(hu3 +k-v3+1w3 )
-I
J3'
(3.16)
The measured intensity is given by IFIl Because of the exponential,
which can take positive and negative values (and complex values), the
diffracted beams from different atoms can cancel each other. One example is a
body-centered cell with only one kind of atom, where one finds that for
h+k+1 even, F = 2f but for h+k+1 odd, F is zero. This means that even though
these reflections are allowed according to Bragg's law, they are not in the X-ray
spectrum. Noting which reflections are missing helps identify the different
structures within the cubic system. If there are different kinds of atoms in the
unit cell, reflections which might cancel each other in a unit cell with only
one kind of atom, might not cancel completely, but still the amplitudes will
be reduced due to the different atomic structure factors.
There are several additional factors which influence the intensities,
like absorption, multiplicity, temperature effects, some of which depend o n
the type of sample or the method used and which are discussed in any basic
text-book on this subject, e.g. Cullity [42].
In summary, the position of the Bragg reflections and possibly missing
reflections gives very basic structural information about the unit cell.
Relative intensities yield information about the atoms in the unit cell, and
are used in refinement techniques to learn about the atomic structure by
49
modeling the structure, calculating the intensities and comparing them to
measured intensities. The width of the peak is influenced by imperfections
like finite crystal size, variations in lattice parameter, caused by strain and
several other possible factors.
No thin film is a true single crystal, but is composed of individual
grains. To gain information about the alignment of the grains, rocking curves
and 0-scans are used.
For a rocking curve, the detector is set at an angle 20 that satisfies the
Bragg condition, and locked in position. The sample is then rotated ("rocked")
through 0, keeping its normal in the sourcedetector plane. This brings
crystallites whose normal to the diffracting plane does not bisect the incoming
and reflected beam in the standard 0-20 set-up into the diffraction condition.
In oriented film samples, the width of a rocking curve is a measure of how
well aligned the different grains of the film are with respect to the normal of
the film surface.
To determine the in-plane alignment, 0-scans are used [43]. In a 0-20
scan, only planes parallel to the sample surface can be detected. If the sample
film surface
diffraction plane
Fig. 3.9 Schematic diagram of Field-Merchant method to determine inplane
alignment of textured samples.
50
is tilted relative to the x-ray beam as in Fig. 3.9, other peaks can be detected. If
for example a film is tilted by a=56.8° (the interplanar angle between the
YBa2Cu3O7 (102) and (001) orientations) and 29 is set to 32.5° (calculated from
(102) lattice spacing), by rotating the sample around the 0-axis, the diffraction
signal will be detectable only if an a-axis oriented grain is parallel to the plane
of incidence of the diffracted x-ray. A plot of 4) versus the diffracted intensity
provides a measure of the distribution of grain orientations in the film
sample. Zheng et al. [44] show that for SrTiO3 and LaA1O3, the a-axis aligns
with the (100) orientation of the substrate, but for MgO, 4% of the grains are
aligned along the (110) direction and in the case of YSZ 66% of the grains are
aligned along the (110) direction. These results are explained by lattice
mismatch, for MgO the mismatch in the (100) direction is
direction
in the (110)
but if one considers the YBa2Cu307_8 [300] and the MgO [220]
directions, the mismatch becomes only 1.8%.
3.4.1 Interpretation of X-Ray Data
The x-ray data is taken on a Siemens D5000 powder diffractometer with
a Kevex Si(Li) detector. It is used as a 0-20 diffractometer, with the added
capability of taking rockingcurves. The spectrometer is set at a current of 30
mA and an accelerating voltage of 50 kV. Appendix B gives the typical
settings of the programs used to investigate the films.
Before any data concerning the film can be extracted from the recorded
spectra, one has to identify which peaks are related to the substrate and which
are film related. Table 3.2 gives the expected 20 positions of the peaks for caxis oriented YBa2Cu307_8 films and Table 3.3 does the same for the substrates
used in this investigation: MgO, LaA1O3 and SrTiO3. Because the substrate
51
contains so much oriented material, the substrate peaks are far stronger than
any of the film peaks. This causes several problems.
One problem is that the strong substrate peaks usually cause the
detector to overload. A peculiarity of the diffraction system used is the fact
that in that case it sets the data to zero as it can be seen in Fig. 3.10. (Because
the detector can be used to determine the energy of incoming photons, it
rejects two photons which come in too close to each other and would give a
Reflection (001)
2 0 [O] (c=11.68 A)
7.56
15.15
22.81
(001)
(002)
(003)
(004)
(005)
(006)
(007)
(008)
(009)
(0010)
(0011)
(0012)
(0013)
30.58
38.49
46.6
54.96
63.66
72.79
82.48
92.97
104.58
117.97
2 0 [0] (c=11.81 A)
7.48
14.99
22.56
30.23
38.05
46.06
54.31
62.88
71.86
81.38
91.65
102.96
115.9
Table 3.2 The expected 0 -28 values for c-axis oriented YBa2Cu306, The left
column is for fully oxygenated material, while the right column is taken from
a powder with an oxygen content of 6.09 [5].
MgO
(100)
(200)
(300)
(400)
42.91
94.04
Sr TiO3
22.74
46.45
72.53
104.13
LaA1O3
23.44
47.95
75.11
108.71
Table 3.3 Expected 29 values in ° for the substrates used in this thesis. MgO is
face centered cubic and does not have (100) and (300) reflections.
52
wrong energy. As a consequence, for part of the substrate peaks the detector
rejects all of the incoming photons.) The next several datapoints after the
overflow do not represent real data either. This means that the substrate
peaks can be used only to give a rough indication of the quality of the
substrate: if there are strong substrate peaks the sample is well oriented. The
reverse is not quite true. Small substrate peaks can indicate a "bad" substrate,
especially in the case of MgO, which can form an amorphous surface layer if
exposed to water vapor. But they might also indicate that the substrate
changer did not pick up the sample correctly or that the sample was not
mounted properly in the substrate holder. Fig. 3.12 illustrates this point very
clearly. First the low intensity spectrum was recorded, then the same sample
was run again, without ever touching or remounting it. The only thing that
touched the sample holder was the sample changer, which picks up the
sample from the loader and puts it into the beam. With low intensity spectra,
the first order of business is to rerun the sample; if that does not help, then
remount the sample. The last step would be running a rocking curve on the
substrate peaks. If the substrate peaks are still small, it is almost certainly an
indication that the substrate is of bad quality.
53
70000
60000
50000
c
0
40000
30000
20000
10000
0
74
74.5
75
76
75.5
76.5
243[1
Fig. 3.10 Overloading of the dectector due to intensely diffracting substrate
peaks. This data was taken on LaA1O3, but the same behavior occurs for MgO
and SrTiO3.
25000
20000
0
15000
Cu Kp line
C
a
0
due to MgO (200)
10000
5000
-
0
37.5
38
38.5
39
39.5
20[°]
Fig. 3.11 Influence of the Cu KR line due to the MgO substrate on the shape of
the YBa2Cu3O7 peak. With a slit with of 0.4°, the spectrometer overflowed and
cut out part of the peak. For a 0.2° slit width, it was able to record the very
high intensity, showing that one sees two peaks. In the 0.1° slit width case, the
Cu KR line does not cause any problems, but the intensity is significantely
reduced. Because the intensity of the substrate peak is still not reduced
enough to avoid the overflow problems, usually a slit width of 0.3° is used.
54
In addition, there can be several lines in the spectra caused by the
substrate diffracting wavelengths other than the intended Cu Ka at 1.54A. In
order to get a reasonable intensity for the Cu Ka line, the electrons are
accelerated towards the Cu target with about 50 kV. This is enough energy to
also excite the Cu KR line. In order to remove this additional line, a 12 gm N i
filter is put into the beam. This filter reduces the intensity of the Ka line to 1%
of the Ka line (and reduces the intensity of the Ka line), but the Ka is still
detectable, because the substrate is so strongly diffracting. This is especially
annoying in the case of the (005) peak for a YBa2Cu307.8 film on MgO, where
the Ka from the substrate (200) peak sits near the YBa2Cu3O7 (005) peak, and is
almost coincident for the 8 = 0 case, as seen in Fig. 3.11.
Further non-film lines may be introduced into the spectrum because of
the tungsten filament used to supply the electrons which are accelerated
25000
00
20000 ­
15000 ­
0
0
0
0
>­
0
0
C/3
>­
0
0
10000
5000 ­
0
38
40
44
46
48
Fig. 3.12 8 -20 spectra on the same YBa2Cu3O7 film deposited on MgO. After the
first spectrum was taken, the spectrum was rerun, without remounting the
sample. The automatic sample changer was the only thing that touched the
sample.
55
towards the Cu target. When the filament is heated up, it evaporates small
amounts of tungsten everywhere in the tube, including on the Cu target.
Because the voltage needed to excite the tungsten L line is relatively low
(8.4kV), there can be some contamination of the spectra with tungsten lines.
Fig. 3.12 shows an example of how tungsten lines will affect a spectrum. The
strength of the tungsten lines depends on the age of the x-ray tube; the older
the tube the more likely it is to see tungsten lines.
I interpret my x-ray data according to the following procedure: I first
check the substrate peaks, and if they overloaded the detector, it means that I
mounted the sample properly. If not, I have to decide if the sample changer
picked up the sample improperly and I just have to rerun it or if I should
remount it.
Then I identify all the YBa2Cu307_5 (001) peaks, checking that they are in
the right positions and that all other peaks can be accounted for by being
substrate related. Any other peaks are an indication of oriented impurities or
other orientations of YBa2Cu307_8.
Identification of peaks that cannot be traced back to the substrate is an
art in itself because in general identification of a material by x-ray diffraction
is done with powders. Powder spectra show all possible orientations and not
just a selected subset as in the case of oriented film samples. The
identification of a powder spectrum is then done by matching the position
and intensities of the observed lines and the measured intensities with
measured spectra in databases like the JCPDF files, which have been created
specifically for this purpose. In oriented film samples, only a few lines are
visible. In addition, if there are two different orientations, their ratio will not
be the same as in a random powder sample, which means that the ratio of
intensities will not be the same as in the data files. The ratios still have to
56
correspond within a family of planes like, for example, the (001) family, but
not in the case of two unrelated orientations. Depending on how many
additional peaks one sees, the identification turns into a little more than
educated guesswork that must be backed by other methods.
Additional information can be extracted from the x-ray spectra by
determining the lattice parameter. This is especially important in the case of
oxygen-deficient films, because even though is it significantly harder to
associate a c-axis length with a certain oxygen content, a change in c-axis
length is an indication that the oxygen content changed.
3.4.2 Determination of Lattice Parameter from X-Ray Data
This following procedure is used to determine the length of the c-axis.
The first step is to find the exact peak positions of all the (001) peaks. This is
done by using a program called PROFILE on Dr. Sleight's x-ray machine,
which fits the peaks from the spectra (see Appendix C). In addition to giving
the peak position, the program also determines the height, area, FWHM and
d-spacing of the selected peak. From this the c-axis length of each (001) peak is
calculated according to
c=
2 sin
(3.17)
To get the most precise lattice parameter measurement, one is
especially interested in the high angle reflections. The reason can be seen by
differentiating Bragg's law with respect to 0.
Ad = -cot 909
d
(3.18)
57
Since cot 8 approaches zero as 8 approaches 90°, the relative error also
approaches zero as 8 approaches 90° or 28 approaches 180° (the maximum
possible angle). Simply calculating the lattice parameters from different
reflections, plotting them against 20 and extrapolating to 20=180° will not
give a very precise result because the resulting function is not linear. Nuffield
[41] and Cu llity [42] show in their respective x-ray diffraction books that if the
calculated lattice parameters are plotted against certain functions of 8, the
resulting curves are linear and can be extrapolated with confidence. Nuffield
especially goes into detail to derive the different sources of error, for example
the displacement of the sample, divergence of the beam and absorption of xrays by the sample and shows that by plotting the calculated lattice parameter
against
cost 9
sine
cost
(3.19)
the most reflections can be used to extrapolate the lattice parameter.
Therefore, the c-axis length is obtained by plotting the c-axis length calculated
from Bragg's law against (3.19) and then extrapolating to zero (which is the
same as letting 8 go to 90°).
For c-axis oriented films, the lattice parameter is typically between
11.7 A for a fully oxygenated film YBa2Cu3O7 and 11.88 A for a YBa2Cu306.25
film. Several factors like cation disorder or inhomgeneity in the sample
influence the c-axis length and the length even for nominally fully
oxygenated YBa2Cu3O7 films with good Tc 's can vary between 11.7A and
11.72A. Data can be found in Chapter 6.
58
Chapter 4
Perturbed Angular Correlation Spectroscopy
The main technique used for analyzing our films is PAC (perturbed
angular correlation spectroscopy). In a nutshell, PAC investigates the
interaction between the quadrupole moment of a probe atom, here mIn, and
the electric field gradient at the site of the probe atom. It is done by measuring
the angular correlation between two 7-quanta emitted during the decay of the
probe atom.
'In substitutes into the YBa2Cu306, structure at one single site. It
decays with a life time of 2.8 days to an excited state of 111Cd. From this state it
decays via a yy cascade into the ground state. The decay scheme for 111In is
shown in Fig 4.1. During the time the probe atom is in the intermediate state,
the quadrupole moment of the probe atom interacts with the electric field
111 In
9+
ti/2=2.83d
2
EC
7+
2
171keV
5+
t1/2=85 ns
2
245keV
1+
V
2
111Cd
Fig. 4.1 Decay scheme for 111In, showing only the transitions relevant for PAC.
59
gradient. This interaction changes the correlation between the first y
quantum and the second one emitted. Compared to the unperturbed case, the
second quantum is emitted in a different direction.
It is also possible to investigate interactions with a magnetic field,
internal as well as external, but in YBa2Cu307_8, the local magnetic field is zero.
We tried one experiment to trap magnetic flux, but only about 10% of the
applied field of 8 T was trapped, which was too small to measure. This thesis
will concentrate only on electric interactions.
4.1 Naive Theory
To get a feel for perturbed angular correlation, one has to look at the
angular distribution of the y-radiation, which is anisotropic in space and
depends on the angle between the angular momentum 1 of the y-radiation
and the quantization axis, commonly referred to as the z-axis. Especially
important is that for certain 1 and m, (the projection of 1 on the z-axis) there is
no emitted radiation along the z-axis.
Fig. 4.2 Emitted radiation pattern for 1= 1 and m = 0,±1 (dipole radiation). For
1= 1, m = 0, there is no emitted radiation along the z-axis.
60
Only by the superposition of all possible states does one get an isotropic
distribution, which means that for all directions one measures the same
intensity. Consider the hypothetical yy cascade with 0-1-0 as the sequence of
spins. If for the first detected 'yquantum the quantization axis is taken as the
line connecting the nucleus with the detector, only quanta with m=±1 can be
detected, because only they have non-zero intensity in the z-direction, as
illustrated in Fig. 4.2. Due to conservation of momentum, this means the
nucleus in the intermediate state (with I = 1) can only have an angular
momentum of M = +1 or M = 1. By choosing a particular quantization axis,
the sublevels are no longer equally populated. If the second quantum is
measured in coincidence with the first one, it can only be detected in the state
m = ±1, because the state M = 0 is not populated. It is important that the
second quantum has to be in coincidence with the first one, for without the
coincidence condition, one would measure an isotropic distribution.
= 0 ; M. =O
1
+1
M = +1
I = 1 ; M = 0
V
M=
+1
1
If= 0 ; Mf = 0
Fig. 4.3 Hypothetical decay scheme of a 0-1-0 cascade. The M levels are
degenerate but are displaced for clarity. Note the population of the M = 0 level
is zero.
61
Up to now we only have a spatial correlation between the two y-
quanta, which can be described by the angular correlation function W(8),
defined later in this chapter. The spatial correlation-was achieved by fixing the
quantization axis by measuring the first y-quantum, causing certain sub­
levels associated with it to become depopulated and then measuring the
second quantum in coincidence with the first. In addition one can start a
counter when the first y-quantum arrives and stop it with the arrival of the
y-second quantum (in a standard set up with four detectors, any detector can
function as a start or stop detector). If there is an interaction between the
aligned spins and some surrounding electric or magnetic field (or both),
either due to the structure itself or an externally applied field, the spins will
start to precess about the axis of the field (in a classical picture) and
subsequently the population of the sublevels will be changed (in a quantum
picture). This in turn changes the angular correlation function W, which now
will depend on 0 and t. This is the basis of PAC-perturbed angular correlation
spectroscopy-; due to the interaction of the intermediate state with some
field, the angular correlation is changed from the unperturbed state. To sum
up the most important points: By measuring the first emitted y-quantum,
certain sublevels for the intermediate state are chosen. Measuring the second
y-quantum in coincidence with the first one leads to a spatial distribution of
the radiation. If the intermediate state is perturbed by some interaction, either
electric or magnetic or both, the emitted radiation does not depend only on
the angle, but also on time. Because in a standard set-up the detectors are fixed
in space, the change of the correlation function with respect to time is what is
being measured.
62
4.2 In-Depth Treatment
Extensive treatment of how to get from the Hamiltonian to the angular
perturbation function W(0,t) can be found in several papers and books. The
most rigorous is by Frauenfelder and Steffen [45]. Additional treatments can
be found in a textbook by Schatz and Weidlich [36] and a paper by Butz [46]. In
order to investigate the interaction between the nuclear electric charge
distribution p(r) and the potential OW caused by the charge distribution due
to the surrounding electrons and positive ions one can write the interaction
energy as
Eelectr =
f p(r)(1)(r)d3r
(4.1)
Expanding the electrical potential in a Taylor series leads to the
following expressions:
E,r = go) ±
+
= 430 jp(r)d
3
r
+i(?43\
a =1
1
p(r)xad'r
(4.2)
axa
020 \
2 a,fl dxadx
p(r)xaxscl3r +
The monopole term E(0) introduces the same energy shift for all
sublevels m for a given nuclear spin I and the dipole term E(1) vanishes. The
quadrupole term E(2), on the other hand introduces an energy shift which
depends on the sublevel quantum number m.
This can be seen more clearly after introducing
a20
(4.3)
dx.dx)3 Jo
43
63
diagonalizing this symmetric matrix and splitting off another monopole term
which describes the interaction of the extended nucleus with the electrons at
the nuclear site. If one additionally defines the electric quadrupole moment as
Q. = -ei p(r)(34,
1-2 )dV?.
(4.4)
and the electric field gradient Vac, via
cl3aa = Vaa
as
(4.5)
one is left with the energy
EQ = -e-6Ict V .Q.
(4.6)
which means that the energy is the product of the electric field gradient with
the nuclear electric quadrupole moment. The m dependence of this energy
comes from expressing the electric quadrupole tensor with the help of
spherical harmonics and calculating the matrix elements (a more in-depth
treatment of the above paragraph is in Ref. [36]).
We take a closer look at the electric field gradient, the quantity of
interest that we measure in the PAC experiments. It is considered an
interaction reaching not much further than next-nearest neighbors and as we
have seen from the above discussion, it yields information about the strength
and symmetries of the charge distribution surrounding the probe atom. Vat; is
a second degree tensor which is symmetric, diagonal and traceless (due to the
definition) and can therefore be described completely by five components.
The Hamiltonian for the quadrupole interaction using the above
definition of the electric field gradient can be written as
HQ =
eQ
6.1"(2I
[3
IV
-(I
1.0+104d-34P]
co
2 a
Criel
(4.7)
64
Because the tensor of the electric field gradient is symmetric, one can
choose a principal-axis system in which Vap = 0 for a # 13. In this coordinate
system, the above equation becomes
HQ
[V. (31. _12)+1(v.
4/(2/
Iy2)]
(4.8)
The above equation shows that two parameters of the electric field
gradient are enough to describe the interaction with the quadrupole moment.
Usually the largest component of the field gradient Vzz and the difference
Vxx Vyy are given. The difference is given by introducing a new parameter,
the asymmetry rb which is defined as
Vxx
=
with I Vxx15_ I Vyy15_ I Vzz I
.
Vyy
(4.9)
Vu
Because the electric field gradient tensor is traceless
and with the above relationship between the
holds. In addition, by
giving Vz, and 11, the other two components can be calculated.
The remaining three parameters which are necessary to describe the
electric field gradient tensor within any system of coordinates are the three
Euler angles. They are used to transform the laboratory coordinate system,
which is usually the host crystal system, into the principal axis system of the
electric field gradient.
4.3 General Perturbation Function
Several intrepid theorists, especially Frauenfelder and Steffen [45] put
the above description into a solid theoretical framework and showed how the
decay of a probe atom via a yy cascade and the interaction of the
intermediate state with either an electric or a magnetic field can be expressed
65
in an terms of an angular correlation function. In its most general form, the
perturbed angular correlation function can be written as
w(pri,pr,,t)
1
Ak (y, )Ak2 (72 )
v(2k2 + 1)
y,Glcv,/N2(t)KAT,;(611,91)17k:2 (02,92)
N1
,N,
(4.10)
the angles and directions are defined in Fig. 4.4, the Yk are the spherical
harmonics, the Aki (y) and Ake (72) are angular correlation coefficients. They
depend on the spins of the nuclear states and the angular momentum 1 of the
emitted y-rays, but are independent of any kind of perturbation. These
coefficients can be calculated for general nuclei (see Ref. [46]) and for the 111In
can be calculated from a table by Ferentz and Rosenzweig (in [45]) and are:
A2(y1) = 0.3299, A2(y2) = -0.5345, Azi(Yi) = 0.0023, A4(y2) = 0.6172
These theoretical values have to be corrected for the finite solid angles of the
detectors and for the distance from the sample to the detector.
Because of conservation of parity, the sum runs only over even ki.
Terms with ki > 4 contribute less than 1% and are neglected. The information
about the interaction of the intermediate nuclear state with spin I is contained
in the Gkr . These perturbation functions are given by:
GrkliN2(t)= I(-1) 2I+ma+ m b 112ki +1112k2 +1 (
ma,mb
I
I
,
ma N
k2
Mb Mb N2) (4.11)
(mb 1A(t), ma )(mblA(t)lm:)*
The A(t) is the time evolution operator, which is given for timeindependent interactions by:
66
A(t) = en`m
(4.12)
14 is the Hamiltonian describing the interaction of the intermediate
nuclear state with its environment via the nuclear dipole or quadrupole
moment.
Butz [46] showed in his paper how the above Gk k'2 can be transformed
into a significantly easier form. This is done by calculating the matrix
elements. Furthermore, it can be shown that for static electric quadrupole
interaction the perturbation functions are real. As a consequence, the
exponentials will turn into cosines during the calculations and one can put
x
Fig. 4.4 Definitions of angles and orientations for detector-probe system. For
clarity, only one detector is shown, but a real PAC set up has at least 2
detectors, the standard set up has 4 detectors, in one plane and at 90° to each
other.
67
the correlation function into the much simpler looking form:
3
W(pri , py, , t)= 1 + So +
S cos con t
(4.13)
n =1
The Sk contain all the information about the orientation of the electric
field gradient and the detectors relative to the crystal lattice, as well as the
angle 0 between pn and 1372. They also depend on the asymmetry 11 as do the
frequencies (Dk. Butz [46] gives analytical solutions for the perturbation
functions and the frequencies, while Wegner [47] gives the perturbation
functions for doped cubic single crystals in different orientations and for
111
cc=0°, 3 =90°
a=45°, 5=90°
cx=0°,13=0°
oc=45°,13=0°
Fig. 4.5 Definition of angles used to describe the orientation of the film
relative to the detectors. a gives the angle between the (100) direction of the
substrate and a line connecting two detectors (at 180° to each other), 13 is the
complement of the angle between the normal to the detector plane and the
substrate normal, when the substrate is rotated about an axis parallel to the
(100) direction. This figure give a bird'seyeview of the four orientations
used in this thesis and the associated a and (3.
68
different values of the asymmetry, stretching over several pages. These
functions are rather complicated and it would go beyond the scope of this
thesis to even start quoting them, because they depend on so many different
factors and they are seldom intuitive.
Because the PAC investigation are carried out on films, the angles
describing the orientation of the substrate relative to the detectors have to be
defined. Positioning the normal of the substrate perpendicular to the detector
plane, a is the angle between the (100) direction of the substrate and a line
connecting two detectors. The angle p is the complement of the angle between
the normal of the detector plane and the normal to the substrate, when the
substrate is rotated about an axis parallel to the (100) direction of the substrate,
starting from the position used to define a. A substrate in the detector plane
therefore has I3 = 90°, a substrate with the normal in the detector plane has p =
0°. Fig. 4.5 shows a birdseye view of the four orientations used in this thesis.
To illustrate how the orientation of the
electric field gradient
influences the spectra, we consider a sample with the main axis of the electric
field gradient pointing perpendicular to the detector plane. There are no
differences (within error bars) between a spectrum taken with the corners of
the film pointing into the detectors (a = 45°) or between the detectors (a = 0°)
(Fig. 4.5 defines these angles).
As a different example, one sees a marked difference if the main axis of
the electric field gradient is in the detector plane. In this case, taking a c-axis
oriented orthorhombic film, which has the main axis of the electric field
gradient along the b-axis, in the a = 0° spectrum, which has the main axis of
the electric field gradient pointing into the detectors, col is the strongest
component, while o)2 is suppressed. In the a = 45° spectrum, with the ma in
axis of the electric field gradient now pointing in between the detectors, w1 is
69
suppressed, while 0)2 is the strongest component. In all orientation, the
amplitude of 0)3 is very small. In a powder spectrum, which averages over all
possible orientations, the amplitudes decrease with increasing frequency.
Fig. 4.6 shows calculations done by Dr. Roland Platzer which illustrate
the influence of the different orientations and the symmetry on the different
1.2
1
a=45°
0.8
/
0.8
0.6
S
Sr.
/
-
0.6
/
0.4
0.4
0.2
0.2
0
0
1
0
20
40
60
80
0
20
I
1
40
13
11
60
I
I
11
80
13
0.8
a=45°.
0.8
0.6
0.6
S
"
S
0.4
0.4
0.2
7 -------------------­
0.2
0
0
0
20
40
60
80
0
r3
20
40
60
80
13
Fig. 4.6 Calculated amplitudes for different orientations, with the electric field
gradient perpendicular to the film plane or in the film plane. The upper two
plots assume the main axis of the electric field gradient perpendicular to the
film plane and ri = 0.4. (Used for c-axis oriented YBa2Cu306..25.) The two lower
plots assume a completely twinned film with the main axis of the electric
field gradient in the film plane and rl = 0.5 (Used for c-axis oriented
YBa2Cu3O7.) The angles are defined in Fig. 4.5.
70
amplitudes and are a summary as well as an extension of the discussion in
the above paragraph. The calculations were done for a c-axis oriented,
tetragonal YBa2Cu306.25 film, with an asymmetry ri = 0.4 as well as for a c-axis
oriented, completely twinned YBa2Cu3O7 film. Note that for the tetragonal
case for 13 = 90°, the Si are almost the same for a = 0° and a = 45°, except for the
hardcore contribution. The main axis of the electric field gradient
is
perpendicular to the detector plane and rotating the film by 45° does not make
much of a difference.
Fig. 4.7 illustrates the above discussion and shows the spectra of an
orthorhombic and a tetragonal film in the same orientation (a = 45°,
-0.15
= 90°).
-0.15
tetragonal film
-0.10
8
8
6
6
4
4
2
2
- -0.10
F
+
200
100
0
300
0
co [Mrad/s]
-0.05
200
100
0
300
[Mrad/s]
-0.05
0.00
0.00
I
100
200
t [ns]
300
400
100
.
200
300
400
t [ns]
Fig. 4.7 R(t) and Fourier transform of a tetragonal (left) and an orthorhombic
film, with the films in the same orientation (a = 45°,(3 = 90°). The R(t) spectra
clearly show the higher frequency of the orthorhombic film, the data
"wiggles" a lot faster. The Fourier transforms shows the orientation
dependence more clearly, with w2 suppresed for the tetragonal film, and
suppresed for the orthorhombic film.
co,
71
All the analysis is done in the time domain, but it is often easier to look at
and interpret the spectra in the frequency domain. As can be clearly seen, in
the orthorhombic phase the amplitude of u is suppressed and in the
tetragonal phase, the amplitude of c02 is suppressed (for the immediate
comparison, one also has to remember that the value of co, in the tetragonal
case is about 1/2 of the orthorhombic case.
The change in the amplitudes for the different phases and orientations
shows clearly that the electric field gradient changes orientation. From our
experiments we deduce that the main axis of the electric field gradient (Vzz)
changes from pointing along the c-axis in the tetragonal phase to pointing
along the b-axis in the orthorhombic case, which is in agreement with the
theoretical calculations by Yu et al. [48].
4.4 Data Acquisition
In order to take a PAC spectrum, at least two detectors are needed.
Several different kinds can be used, depending on requirements for energy
and time resolution. We use BaF2 scintillation detectors (At =300 ps for 511
keV yrays). The yquantum enters the detectors and interacts with the
detector material either due to the photoeffect or Compton effect, creating an
excited electron. This electrons excite (or ionize) atoms in the scintillator with
the number being proportional to Ee. The excited atoms relax by emission of
light, which enters a photomultiplier that gives a signal proportional to the
energy of the original yquantum. One detector is used as the start detector,
and starts a clock when yi is detected. The second one stops the clock as soon
as it detects y2. For 'In, yi has an energy of 171 keV and y2 of 245 keV, and
the detectors can clearly differentiate between the two 1quanta. In a standard
72
4detector set up, each detector can be a start and stop detector. The time
difference t are binned by a multi-channel analyzer. Over time, a coincidence
spectrum is collected.
The coincident count rate Nib between a start detector i and a stop
detector j is related to the angular correlation function W(0,t) by:
Nv(0,t)=Noexp(---t---)W(9,t)+B
(4.14)
TN
B is the time-independent background random coincident count rate
and is subtracted, TN is the lifetime of the intermediate level (85 ns for 111In).
In order to collect as much information as possible, several detectors
are used with the most common arrangement being four detectors at 90° to
each other, each of which can be used as a start or stop detector. This leads to
12 possible coincidence spectra. In order to extract the perturbation function
W(0,t) and hence the asymmetry and the orientation of the electric field
gradient, different count rate ratios can be formed. The ratio used here is:
R(t)= 2
N13(180°,0+ N24 (1 80°, t)
2
No (180°, t)N24(180°,t)
2N23(90°,t)
3
N14 (90 °, t)N23 (90 °, t)
NI 4(90°, t)
(4.15)
The second ratio is a very close approximation of the first one and shows
more clearly that the detector efficiencies cancel, if self-absorption in the
sample is negligible [49], which is the case in a thin film.
This equation is the starting point for extracting data. The above
equation is fitted to
R(t) = A22 Efit S, (n)coskon(n)tiexp(Acont)
(4.16)
n=0
A22 is the effective anisotropy, typically = 0.1 for 111In. In the case of
several different probe sites, an appropriate fraction fi is assigned to the
different sites with If = 1. The S, contain the information about the
orientation of the electric field gradient, while the con contain the information
73
about the asymmetry ri and the strength of the electric field gradient Vzz. The
exponential term is included to account for minor variations in the electric
field gradient due to impurities and other defects.
To get as good a fit as possible, good statistics are crucial. Spectra should
have at least 20000 counts, but ideal would be 100000. Right after deposition,
about 2000 counts/hour are accumulated, but even at this count rate, it would
take over two days to get 100000 counts. We make do with the 20000 counts in
order to be able to perform as many experiments as possible.
4.5 Data Analysis
The actual fitting is done by a program written by Dr. B. Lindgren
(Uppsala University, Sweden) and modified by Dr. Roland Platzer. The
program uses a least squares fit to vary the frequencies, asymmetry, the
frequency distribution, the fraction of probes in one site as well as the Euler
angles (and therefore the orientation of the electric field gradient), trying to
achieve the best fit possible, with the user determining which parameters
should be fit parameters and which should be fixed. It takes the data and an
estimate for the Euler angles which connect the laboratory frame with the
principal axis system of the electric field gradient. (Even though both
coordinate systems use x, y and z for their axes, none of these axis have to be
the same (but they can be)). The lab system is transformed into the principal
axis system of the field-gradient via three rotations characterized by these
Euler angles.
The fitting program calculates R(t) from equations (4.10) and (4.11). and
compares it with the measured R(t). Calculating this R(t) is the heart and soul
74
of the program. Equation (4.11) contains information about the transition
itself, while equation (4.10) contains the information about the orientation of
the electric field gradient due to the spherical harmonics, which are tied to the
laboratory coordinate system via the Euler angles.
Even though it sounds fairly automatic, there are still points were
human intervention is needed. Because the program does the fit numerically,
step sizes are involved. This can become important especially with the Euler
angles, because one has to differentiate them numerically, and it is possible to
overshoot with too big a stepsize. An additional factor, especially in films
which are inherently more disordered than crystals, is the so-called missing
fraction, from probes which are not in a well-defined surrounding and
therefore only contribute to an amorphous background signal. In our films
this fraction is about 40%, and must be taken into account, to achieve a good
fit. To achieve the best fit, several spectra for different orientations are fitted at
the same time. The best fit with respect to the Euler angles is achieved with
fitting the a = 45° spectra simultaneously.
The user also has to determine the number of parameters to be fitted
and check if the results obtained are reasonable or whether one should start
with a different set of initial parameters. (As can happen with fitting routines,
if the starting parameters are too far away from the true values, the routine
might not find the solution). Also, depending on what orientations have
been measured, not all variables can be fitted. For a tetragonal, c-axis oriented
YBa2Cu306.25 film,
ri cannot be determined if the sample has only been
measured in the p = 90° (horizontal) orientation, because in this case, the
main axis of the electric field gradient is perpendicular to the detector plane.
(This is not quite true; for samples with a narrower linewidth is would be
possible to fit the
but we need two orientations to get a good fit.)
75
Some comments have to be made with respect to the fast Fourier
transforms (FFT) commonly shown in the spectra which are intended only as
a guide to the eye. All of the analysis is done on the R(t) spectra, but it is easier
to see which component is the strongest in the Fourier spectra. Because the
time window is limited to 500 ns (and the function has to converge to zero for
the FFT to work, which is done by subtracting the background and using an
envelope function), there already is an uncertainty of 2 MHz in Fourier
spectra, which for the 20 MHz tetragonal line is 10%. This linewidth is not the
true linewidth, which is calculated from the fit done according to (4.16). If one
of two lines looks narrower in the FFT, the linewidth will also be narrower in
the fit.
The linewidth may depend on the temperature. It first increases with
increasing temperature, because with the motion of the atoms due to the
temperature, each probe atom experiences a slightly different environment,
which translates into a broadened linewidth. Once the timescale of this
motion is of the same order as the lifetime of the intermediate state or faster
the motion averages out, the probes all see an average environment and the
linewidth narrows again. This is referred to as motional narrowing.
4.6 Perturbed Angular Correlation Experiments in YBa2Cu3078
Nuclear hyperfine spectroscopies are among the myriad techniques
brought to bear on YBa2Cu307. Nuclear magnetic resonance and Mossbauer
Spectroscopy are more common techniques and their results are nicely
reviewed in Ref. [50], but PAC has the advantage of being more sensitive and
temperature independent. PAC measures the strength, and in oriented
material, the direction of the principal components of the electric field
76
gradient. Virtually all the PAC work on YBa2Cu307_8 has been done on
powders, because it is easier to incorporate the radioactive isotope. However,
in powder or polycrystalline samples, information is' lost about the
orientation of the electric field gradient with respect to the crystal structure,
because the microcrystallites are randomly oriented.
PAC in YBa2Cu307_,5 poses several problems. The first is how to get the
probes atoms into the material, because fairly stringent criteria have to be met
for an atom to be suitable for PAC. Usually 111In is used, even though it is not
part of the YBa2Cu307_45 structure, but 133Ba is not a very favorable probe atom
[51] because of its small anisotropy and short halflife of the intermediate state.
There are several different ways to introduce mIn into YBa2Cu307_5, each with
its own limitations. Bartos and Uhrmacher [52] introduced the PAC probes
into pressed powder samples via ion implantation, accelerating the In ions to
energies of 400keV, achieving an implantation depth of .400 nm. This
method has the disadvantage that it causes radiation damage, clearly visible
in their PAC spectra, and has to be annealed out, using temperatures between
975K and 1275K and another annealing step at 775K in order to restore the
oxygen content. A different approach was chosen by Saitovitch et al. [53], who
diffused the indium into their powder samples at temperatures between
500 °C and 750 °C. Schwenker [54] and Fiissel [55] used a sol-gel method to
incorporate the indium during the formation of the YBa2Cu307_8 powder.
The approach taken in our laboratory is to implement a fourth source
in the evaporation system [56] which incorporates about 1011 atoms during the
deposition of the film, causing the indium atoms to be incorporated while the
structure is forming.
In most of the early 111In:YBa2Cu30,8 PAC experiments, several
different sites were identified, most attributed to 'In substitution at different
77
sites within the YBa2Cu307_5 structure. Bartos and Uhrmacher [52] have an
extensive list in their paper summarizing the frequencies and the suggested
site identifications.
More recent work on PAC on YBa2Cu30,_8 and possible impurities in
'In substitutes at only one location in the
YBa2Cu30,_6 lattice, with all the evidence pointing towards the yttrium site
Refs. [55], [57-61] show that
with a vQ = 39 MHz, and ri = 0.5. All the other observed sites can be attributed
to impurities, formed either during the sample preparation or while
incorporating the indium into the structure and the anneal that usually
follows.
The evidence for the indium substituting for yttrium is only
circumstantial, but several good arguments can be made. Both indium and
yttrium have valence 3+ and their ionic radii are very close (ry = 0.89A,
rh, = 0.91A), while the radii for Cu and Ba are 0.69A and 1.35A, respectively. A
study done by Weidlich [62] showed that up to 3 atom % of the yttrium can be
exchanged with indium. If doped above the solubility limit, InBa2Cu30
forms. Uhrmacher and Bartos [58] show that using a simple point charge
model to calculate the frequencies, the yttrium site has a significantly lower
frequencies than all the other sites. Furthermore, the PAC data taken in this
laboratory [27] agree very well with the theoretical predictions by Yu et al. [48]
for a substitution at the yttrium site. Additional work by me, presented in
Chapter 5, shows that if an impurity can be identified in the x-ray spectra, only
the impurities containing yttrium will show additional PAC signals. None of
these arguments taken by itself is strong enough to state clearly that the
indium substitutes at the yttrium site, but taken together, they make a very
strong case.
78
Measuring the electric field gradient at the yttrium site is especially
important because the two main theories by AmbroschDraxl et al. [63] and by
Yu et al. [48], while agreeing for all the other possible sites in the orientation
and the strength, differ for the yttrium site. Yu et al. [48] calculate the main
axis of the field gradient to be pointing along the b-axis with a Vzz of
0.201x1022Vm-2 and an asymmetry
= 0.43, the AmbroschDraxl [63] paper
calculates the main axis of the electric field gradient pointing along the c-axis
with a Vzz of 0.34x1022Vm-2 and an asymmetry r1=0.9. Because all the other
research had been done on powders', the orientation could not be
determined. Tom et al. [56] found that in twinned YBa2Cu307_,5 films the
electric field gradient was in the ab-plane, but the research on a-axis oriented
films in this thesis was needed to determine the direction of the main axis of
the electric field gradient, which points along the b-axis, as predicted by Yu et
al. [48].
Observing the electric field gradient can be used to get some insight
into the motion of oxygen in YBa2Cu307_5. The frequency decreases with
decreasing oxygen content as shown by Plank et al. [64] or with increasing
temperature (which according to the phase diagram is essentially the same as
decreasing the oxygen content [65]). In addition, work done here on tetragonal
films (Ref. [65] and this thesis) showed that with the phase transition the
main axis of the electric field gradient changed from being along the b-axis to
being along the c-axis. Troger and Butz [51] used 133Ba as the probe atom which
has the advantage of being very close to the Cu-0 chains which are
responsible for the oxygen ordering, found a static electric field gradient for
12 Plank et al. [64] published a spectrum done on a YBa2Cu307_5 film on SrTiO3,
but the spectrum is more likely to be of Y2Cu2O5. They identified it as the
Cu(1) site in YBa2Cu307_6 and did not determine the orientation of the main
axis of the electric field gradient.
79
temperatures up to 400 K (100 °C). For higher temperatures the observed data
can be explained by a fluctuating electric field gradient due to oxygen
diffusion. Theoretically, their data could be used to derive a hopping rate, but
because of insufficient accuracy in the data they did not try this approach.
80
Chapter 5
a-Axis Oriented YBa2Cu306,x Films
YBa2Cu306, can be grown epitaxially in several preferred orientations.
The most commonly grown orientations are with the c-axis normal to the
substrate or the a-axis normal to substrate. It is usually easy to grow c-axis
oriented material with very good characteristics, but for films with the a-axis
normal to the substrate, the characteristics are not quite as good as in general
they have lower T but there is interest in growing films in this orientation.
From an applications point of view, it is advantageous to grow films
with the a-axis normal to the substrate because the current transport takes
place mostly in the ab-plane. The coherence length along the ab-plane
nm) [66] is much longer than the coherence length along the caxis (47-0.1-0.3 nm), which makes c-axis oriented films less suited for the
production of superconductornormal conductorsuperconductor (SNS) or
superconductorinsulatorsuperconductor (SIS) Josephson tunnel junctions.
There is another reason to investigate the a-axis films. Previous PAC
experiments in this laboratory on twinned c-axis oriented films were able to
establish that the main axis of the electric field gradient in YBa2Cu3O7 is in the
ab-plane but, due to the twinning, could not distinguish between the a- or b-
directions. Furthermore, the two main theories differ in the predicted
orientation of the main axis of the electric field gradient with respect to the
yttrium site. Yu et al. [48] calculated the main axis of the electric field gradient
at this site to be pointing along the b -axis, but AmbroschDraxl [63] calculated
it as pointing along the c-axis. Our PAC experiments on a-axis oriented films
were able to confirm Yu's predictions.
81
The usual consensus is that in order to grow a-axis oriented films, the
temperature of the substrate has to be about 100 °C 13 lower than in the case of
the c-axis film (see for example Refs. [67-69]). Yet there are several other
factors which also play an important role in what fraction of the film grows in
which direction. The choice of substrate is important and the optimum
temperature varies for different substrates, a fact that Ref. [70] demonstrates
very clearly. They conclude that a-axis growth is favored by good lattice
matching and in order to achieve
c-axis growth,
a higher substrate
temperature is needed. In addition, the oxygen pressure during the deposition
influences the orientation, shown in Refs. [71] or [721. Terashima et al.
[71]
investigated films sputtered at temperatures between 530 °C and 680 °C. At an
oxygen pressure of 0.032 Pa there was only c-axis oriented growth, while for
pressures greater than 0.65 Pa, only a-axis oriented growth was found. On the
other hand, sputtering a film at 730 °C yielded c-axis oriented growth even at
a pressure of 0.32 Pa. Zhong et al. [72] found, using MOCVD to deposit their
films, that the higher the total partial pressure of the oxidizers, the wider the
temperature window they had to grow films in a purely a-axis normal
orientation. Yet all these papers show that the most important factor is
lowering the substrate temperature, the other factors listed above only
improve the ratio of a-axis oriented grains to c-axis oriented grains.
On the other hand, a lower substrate temperature during deposition is
detrimental to the quality of the film produced. a-axis films usually show
reduced Tc and their resistance vs temperature curves do not have the shape
that is expected from a good c-axis film, namely linear (extrapolating to zero)
13 The different growth techniques like sputtering, MBE, reactive co­
evaporation or laser ablation have different optimum temperatures for the
respective orientations. This is due to the fact that, depending on the
technique, the evaporants hit the substrate with different energies.
82
until slightly higher than the transition temperature, with a very narrow
transition (< 1 K). Raven et al. [73] show nicely how the shape of the
resistance vs temperature changes with the change in the ratio of a-axis
oriented grains to c-axis oriented grains. With increasing a-axis fraction, the
transition temperature is lower, the curves remain linear, but the slope
becomes shallower. One of their resistance vs temperature curves first shows
a slight decrease in resistance with decreasing temperature, but at around
200 K, the resistance rises again, before the sample goes through the
superconducting transition. They also show that this change in the resistance
curves is different from the change introduced by a change in oxygen content,
which changes the transition temperature, but not the slope of the curve. Li et
al.
[69] investigated the dependence of several film characteristics with
decreasing substrate temperature. The T, of a film deposited at 690 °C, which
does not show any a-axis growth, is 90 K, decreasing with decreasing substrate
temperature so that for the film deposited at 570 °C, which clearly shows aaxis peaks in the x-ray diffraction spectrum, the T, is 82 K. The critical current
density decreases from being higher than 106 A/ cm' for the film deposited at
650 °C to slightly more than 104 A /cm2 for the film deposited at 570 °C. The
FWHM (full width half maximum) of the (006) x-ray diffraction peak
widened from about 0.3° for the film deposited at 690 °C to almost 0.6° for the
film deposited at 570 °C, indicating decreased crystallinity with increasing aaxis content.
In order to improve the quality of the films, a template method, with
either the same material or a different one, is often used. This technique was
pioneered by Inam et al. [74]. Trying to grow YBa2Cu307_8/PrBa2Cu307_y14
'4 PrBa2Cu30,_y is not superconducting and is used in conjunction with
YBa2Cu307_8 to make devices with several epitaxial layers
83
superlattices, they noted that the PrBa2Cu307_y layer occasionally had grains
with the a-axis normal to the substrate and that this was less common in the
YBa2Cu30,_5 layer. YBa2Cu307_8 that nucleated on top of these a-axis oriented
grains grew with an a-axis orientation under conditions that normally yield
c-
axis oriented material. Investigating the optimum conditions, they showed by
ion channeling experiments that there is an optimum temperature for c-axis
oriented growth (820 °C) and an optimum temperature for a-axis oriented
growth (700 °C) of the template layer with the region in between showing
mixed a- and c-axis growth and decreased crystallinity. YBa2Cu307_8 films
deposited at 800 °C on a template layer of PrBa2Cu307_y deposited at 700 °C
showed less than 5% c-axis oriented grains in x-ray diffraction studies and a Tc
of 83 K in ac susceptibility measurements. A more recent paper by Trajanovic
et al. [75] using the same technique achieved Tc's of up to 88 K. Rosova et al.
[76] showed that the deposition temperature of the template influences the
volume fraction in the YBa2Cu306, film with 98.6% for a template deposited
at 680 °C, decreasing to 56.1% for a template deposited at 700 °C, even though
the films were all deposited at the same temperature of 710 °C. This result can
be interpreted with the help of the paper by Inam et al. [74], because it clearly
shows that the amount of a- and c-axis oriented grains in the template layer
determine the orientation in the YBa2Cu307_8 film.
Sodtke and Miinder [77] investigated the reason for the Tc suppression
for films deposited with a PrBa2Cu307_y template with Raman spectroscopy
and found that the frequencies, especially of the 500 cm' phonon vibration,
do not indicate any oxygen deficiency, but that the FWHM narrows with
increasing deposition temperature of the film (the template being deposited at
the same temperature for all the films), especially for the apical oxygen (the
0(4) atom, compare Fig. 1.2) and the plane copper vibration modes. This
84
indicates that during the growth at lower temperatures, a very stable disorder
is introduced among the chain oxygen and the apex oxygen. If the deposition
temperature is raised, this disorder vanishes and consequently the Tc rises.
Another method is the so-called self-template method. In this case, a
template of YBa2Cu306, is deposited at a lower substrate temperature, again
resulting in an a-axis film of reduced quality and then the temperature is
raised. The template forces the film to remain in the a-axis orientation despite
the higher temperature, but the higher deposition temperature results in
better film characteristics. Hamet et al. [78] raised the temperature to that
normally used to deposit c-axis oriented films without stopping the
deposition, which seems to be the most common way. Ha et al. [79] used a
self-template method, stopping the deposition between the deposition of the
template layer and the YBa2Cu306, film. Mahajan et al. [80], in a very
thorough paper investigating the influence of several deposition parameters
in sputtering using the self-template layer, found that stopping the deposition
while raising the temperature did not change the Tc and ATc for films
deposited on MgO and SrTiO3, but stopping the deposition while raising the
temperature introduced c-axis and (110)-oriented grains in addition to the a-
axis oriented grains for films deposited on MgO (and another perovskite
phase which they mentioned, but could not identify). Yet if the deposition
was not stopped during the raising of temperature, only a-axis oriented grains
and the unidentified phase were observed. On the other hand, films
deposited on SrTiO3 and NdGaO3 showed only a-axis oriented grains in both
cases. They explain the results on MgO by comparing it to "the seeding process
in crystal growth in which the growth of a crystal occurs by slow pulling of a
seed crystal with the purity and properties of the resulting crystal depending
on the pulling rate."
85
A question addressed in a paper by Kawamoto and Hirabayashi [81] is
why the films grow with only the a-axis perpendicular to the substrate rather
than the b-axis perpendicular to the substrate. The question is important,
because it is related to oxygen ordering. For a-axis oriented grains, the Cu-0
chains are parallel to the substrate, while for the b-axis oriented grains, the
chains run perpendicular to the substrate surface. Their explanation considers
the oxygendeficient state in which the films are grown. During the
subsequent oxygen loading, it is energetically preferable to form chains
parallel to the surface, which corresponds to a-axis oriented material, because
there are fewer terminations for the Cu-0 chains than for chains running
perpendicular to the substrate surface. Citing a paper by Ronay and
Nordlander [82], who calculated the activation for the oxygen motion
(energies of 1.7 and 1.6 eV are required for motion in the a- and c- directions
respectively and no energy barrier in the b-direction) they conclude further
that once the short Cu-0 chains parallel to the surface have formed, they will
grow very fast.
5.1 X-Ray and Resistance Studies on a-Axis Oriented Films
Before doing PAC experiments on a-axis oriented films, it was
necessary to characterize their growth mode. Especially important is the effect
of the substrate temperature and local oxygen pressure at deposition.
A series of films at decreasing substrate temperatures was grown,
starting at 670 °C, which had proved to be the best temperature to grow high
quality c-axis oriented films on SrTiO3, all the way down to 570°C, which is
100°C lower. Fig. 5.1 shows the development of the YBa2Cu306, (200) peak,
86
400
-T =670°C
d
300
Td=650°C
T =610°C
d
Td=600°C
200
- T =590°C
d
100
...
LaA103(200)
-.
-100
46
46.5
47
47.5
48
48.5
49
2 0 [°]
Fig. 5.1 Rise in the (200) peak of YBa2Cu3O7 with decreasing deposition
temperature. All films were deposited at 25% of 50sccm of oxygen (included is
data from a film deposited at 590 °C, where the peak height decreases again).
The x-ray data is normalized with respect to the (005) peak. The position of
the LaA1O3 (200) is indicated; the data are suppressed because substrate peaks
cause the detector to overload.
which clearly increases with decreasing temperature until 600 °C, after which
it decreases.
The 0-20 scans taken at OSU, as well as pole-figures taken by Dr.
Roberta Bigelow at Willamette University, were carefully checked for any
evidence of b-axis oriented grains. If there are any b-axis oriented grains, they
are below the detection limit of x-ray diffraction, which is usually around 5%.
Table 5.4 shows that the separation of the (n00) and (OnO) peaks, caused by the
orthorhombic distortion is at least 0.4°. The peaks should be very clearly
resolved for n = 3 and n = 4 (the linewidth of the peaks is less than 1°, as
shown in chapter 3).
87
n
LaA1O3 (n00)
1
23.444
23.258
22.833
2
47.949
47.549
46.643
3
75.107
74.416
72.859
4
108.714
107.469
104.703
YBa2Cu307 (n00)
YBa2Cu307 (0n0)
Table 5.4. 29 (in °) values for LaA1O3 and a-axis and b-axis oriented YBa2Cu307.
LaA1O3 has a lattice constant of 3.79A, the a-axis of YBa2Cu307 is 3.82A, the baxis of YBa2Cu307 is 3.89A.
It is difficult to quantify the ratio of a-axis to c-axis oriented grains from
the x-ray diffraction spectrum, but it is useful to compare the maximum peak
intensities of the (005) peak, representing the c-axis grains, and the (200) peak,
representing the a-axis grains (Fig. 5.2). The ratio is used because the absolute
number of counts is not fixed. Fig. 5.2 shows a maximum in the a-axis content
at about 600 °C. Because the 0- values for the (h00) of YBa2Cu30,6 and LaAIO3
are very close together, as shown in Table 5.4 and Fig. 5.1, the substrate
contributes significantly to the peak intensity of the (200) peak. Rocking
curves on the (200) and (005) peak of the YBa2Cu307 were performed to
minimize the substrate interference. Because of the slightly different 0 values
required to bring the substrate and film into the diffraction condition (Table
5.4) the substrate will contribute only minimally, as seen in Fig.
5.3
Integrating the area under these curves and taking the ratio gives an estimate
of the amount of material in each of the orientations.
88
350
300
250
200
150
100
50
0
540
560
580
600
620
640
660
680
T substrate"­r°C]
Fig. 5.2 Ratio of the maximum peak intensities of the (200) and (005) peaks.
All films were deposited at 25% of 50 sccm of flowing oxygen, which is a local
pressure at the substrate of 1.8x10-3 torr.
30000
25000
20000
(n
t.
z0C
15000
0
10000
5000
0
21
22
23
24
25
26
27
8
Fig. 5.3 Rocking curves of an a-axis YBa2Cu306, film on a LaA1O3 substrate
and the LaA1O3 alone. The diffractometer was set at 20 = 47.549° for both. The
substrate curve was obtained by flipping the substrate over so that the film did
not contribute. The film was deposited at 590 °C and 15% of 50 sccm.
89
The previous set of films was deposited at various substrate
temperatures at 25% of 50 sccm of flowing oxygen during deposition, which
corresponds to a pressure of 1.8 mtorr at the substrate (Fig. 5.4). During the
deposition, oxygen is supplied at the substrate and therefore the local pressure
at the substrate is significantly higher than the background pressure (typically
torr).
A similar investigation into the pressure dependence of the growth
mode was carried out for films deposited at a temperature of 600 °C, the
temperature of the maximum a-axis content found for films deposited at 25%
of 50 sccm. The result is shown in Fig. 5.5, which indicates that the lower
pressure of 1.1 mtorr at 600 °C optimizes a-axis growth.
90
100
80
CZ'
0
0
60
0
0
40
20
0
540
560
580
600
T substrate
620
l
640
660
680
Fig. 5.4 Ratios of the integrated intensities of rocking curves on the (200) and
(005) peaks. All films were deposited at 25% of 50 sccm of flowing oxygen.
There is a maxium for a deposition temperature of 600 °C.
local pressur [tom]
1
le
1.5 10-3
2 103
2.5 10-3
600
500
0
400
300
(NI
200
100
0
5
10
15
20
25
% of oxygen
30
35
40
Fig. 5.5 Dependence of the ratio of the integrated intensities of the (200) and
(005) peak (calculated from rocking curves) of YBa2Cu3O7 on the pressure
during deposition. All films were deposited at a substrate temperature of
600 °C.
91
The all-important question of T, is addressed in Fig. 5.6, which shows
some of the Tc's for a series of films, deposited at decreasing temperatures
with an oxygen pressure of 25% of 50 sccm (1.8 mtorr). The slope flattens and
for films deposited at relatively low deposition temperatures, the resistance
rises before the films go through the transitions. The decrease in the
transition temperature T, and the increase in the width of the transition is
illustrated in Fig. 5.7. Neither is surprising, because the films were grown
without a template15, which generally improves the crystallinity, as shown in
the literature overview. In the literature, a reduced T, is in general attributed
either to disorder within the oxygen sublattice or to cation disorder, both of
which are more likely to occur at reduced substrate temperature, which
reduces diffusion in the newly deposited film.
15 In our system, it is not practical to grow a template layer, because of the
short deposition time. Shuttering the substrate, changing the substrate
temperature, waiting for the heater to stabilize and than continuing the
deposition requires more material in the boat, but the deposition rates,
especially of the barium tend to be less stable if the boat is overfilled.
92
11111111IIIIIIIIIIIIIIIiI1
1
---
0.8
U)
rn
0.6
CC
0.4
CC
= 670 °C
substrate
T
0.2
= 650°C
substrate
= 590°C
substrate
0
1111111111_111111111111
0
50
100
150
200
.1111
250
300
T(K)
5.6 Normalized resistance for several films grown at decreasing
deposition temperatures with increasing a-axis content. The pressure during
the deposition was 25% of 50 sccm. It is important to note that the slope and
the general shape change, becoming increasingly flatter. This is especially
important to note in comparison with the oxygendeficient films.
Fig.
Comparison of Fig. 5.7 and Fig. 5.8 shows that an optimum in one
characteristic
of the film does not necessarily mean that the other
characteristics are optimized under the same conditions. Fig. 5.8 shows clearly
that at the maximum a-axis content (15% of 50 sccm), the T, is substantially
reduced.
From Fig. 5.7, we see that T, can be recovered by increasing the oxygen
pressure at deposition. However the maximum a-axis concentration for this
pressure is 6 times lower than for the pressure shown in Fig. 5.8.
93
86
13
12
84
11
10
82
T[K]
9 AT c[K]
80
8
7
78
6
76
560
5
580
600
620
640
660
680
Tsubstraterl
Fig. 5.7 Development of T, and AT, with decreasing substrate temperature.
Both show clear evidence of increasing disorder, the T, decreases and the AT,
increases with decreasing deposition temperature (films were deposited at
25% of 50 sccm (1.8 mtorr at substrate)). The vertical arrow indicates the
temperature where a maximum in the amount of a-axis grains was found.
75
70
[K] 65
60
55
5
10
15
20
25
30
35
% of 50 sccm
Fig. 5.8 Dependence of T, on the oxygen pressure for films deposited at 600 °C.
Fig. 5.5 shows that the maximum amount of a-axis oriented grains is for an
oxygen pressure of 15% of 50 sccm, but for this value the T, is significantly
reduced.
94
5.2 Investigation of a-Axis Oriented Films with PAC
The preliminary investigation showed that it was possible to grow aaxis oriented films in our evaporation system. Several a-axis oriented films
were made, with an oxygen content of x = 7 as well as x = 6.25. The fully
oxygenated films were used to establish the direction of the electric field
gradient in YBa2Cu3O7, the YBa2Cu306.25 films were grown to further
investigate the tetragonal phase, especially because at that time the second site
"mystery" had not been solved. Because of the known orientation of the
electric field gradient for the tetragonal phase, they could also be used as
-0.15
-0.10
U
1.00
a =0°, R= 90° ­
0.50
0.75
1
-0.05
S
02
0.25
0.00
-0.15
JN. 0.00
-0.10
0.75
-0.05
a=45 °, fl = 90°
0.50
0.25
0.00
0
100
200
300
time [nsec]
400
0
100
200
0.00
300
[Mrad/s]
Fig. 5.9 R(t) and Fourier transforms of an orthorhombic a-axis oriented
YBa2Cu3O7 film at room temperature. For a c-axis oriented film in the upper
(a=0°, (3=90°) orientation, o), would have the higher amplitude and 632 would
be suppressed. For the lower orientation (a=45°, 13=900), a c-axis oriented film
would have col suppressed and 0)2 would have the higher amplitude.
95
another test that the films were a-axis oriented. (This proved to be the only
use, because for some reason, the film became disordered in the PAC signal if
annealed in oxygen at rather low temperatures
200 °C). Surprisingly they
still showed x-ray peaks.) The PAC a-axis oriented films were grown at 580 °C,
and an oxygen flow of 25% of 50 sccm, not quite at the optimum conditions
shown earlier in this chapter.
5.9 shows the typical PAC spectrum of an a-axis oriented
orthorhombic YBa2Cu3O7 film. In comparison to a c-axis oriented film
Fig.
(compare Fig. 4.7) one sees that for the same orientation of the film, col is
suppressed in the a-axis case and (02 is suppressed in the c-axis case.
-0.15
a=45°,f90° -0.10
II
MR1111.
7
6
5
4
-0.05
3
2
0.00
1
-0.15
0
7
6
-0.10
5
4
-0.05
3
2
0.00
1
0
100
200
time [ns]
300
400
0
100
200
300
400
0
500
co [Mrad/s]
Fig. 5.10 R(t) and Fourier transform of a tetragonal a-axis oriented
YBa2Cu306 25 film. For a tetragonal c-axis oriented YBa2Cu306.25 film, these two
spectra are nearly identical.
96
Fig. 5.10 shows the spectrum of a tetragonal a-axis oriented film with
the plane of the film in the detector plane. These spectra very clearly show the
difference between the a-axis and c-axis oriented films in this particular
orientation. For a c-axis oriented films, these two spectra should be nearly
identical, because the main axis of the electric field gradient is perpendicular
to the detector and film plane and turning the film by 45° in the detector
plane does not make a difference. For the a-axis oriented films, on the other
hand, the main axis of the electric field gradient is in the film and detector
plane. For a = 0°, the main axis of the electric field gradient points into the
detectors, while for a = 45°, the main axis of the electric field gradient points
between the detectors, so one expects to see a significant difference in the
spectra.
The most important result from the a-axis films is that we finally could
determine the complete orientation of the electric field gradient at the
yttrium site. As predicted by Yu et al. [48], it points along the b-axis, with
Vzz = 2.1(1)x10' Wan' and an asymmetry T1= 0.5. (The predicted values are
Vzz = 2.01x10' V/cm2 and
= 0.43.) This fit could only be obtained if a b-axis
fraction of about 15%-20% was assumed, a number which was not confirmed
by x-ray diffraction. These differences have not been resolved yet, but one
educated guess can be based on the fact that x-ray diffraction and PAC measure
on two very different length scales. PAC does not measure much farther than
nextnearest neighbors (20A to be generous), and very locally the structure
could be b-axis oriented, maybe even induced due to the foreign 1111n atom
(which is after all an impurity) but not more than a couple of unit cells in
either direction. X-ray diffraction would not be able to see such small domains
(it averages over at least 1000A) and would only show the a-axis oriented
material. The probe atom locally changing the structure from a-axis to b-axis
97
would also go a long way towards explaining the fairly high number of 20 %,
because the probe atoms are in a layer only about 500A thick (in a 2000A thick
film), but x-ray diffraction averages over the whole thickness of the film.
98
Chapter 6
Oxygen-Deficient Films
The investigation in this thesis was started by the fact that we could
easily produce films with an oxygen content of = 7 and with an oxygen
content of 6.25. From our experiments using PAC we knew that in the case of
YBa2Cu307_6 the main axis of the electric field gradient was pointing along the
b-axis with a frequency vQ = 39 MHz, in the case of YBa2Cu306.25 the main axis
was pointing along the c-axis with a frequency vQ = 20 MHz. Yet we did not
know what happened in between this two values, for example in what way
the frequency changes with respect to the oxygen content and for what oxygen
content the orientation of the electric field gradient changes.
6.1 Investigation of Tetragonal YI3a2Cu306..
The tetragonal phase of YBa2Cu306, shows semiconducting behavior
and not
the
superconducting behavior
which
is
the hallmark
of
orthorhombic YBa2Cu307. There are several good reasons for studying the
tetragonal phase. The main reason is that films are deposited at high
temperatures and low pressures. Low pressures are needed in order to use
vacuum deposition techniques successfully, and it is common knowledge
that in order to produce high quality thin films, high substrate temperatures
during the deposition are needed. If the films are deposited on "cold"
substrates, they are amorphous and a post-anneal at temperatures higher than
700 °C is required to induce crystallinity [83-85]. Low pressures and high
temperatures put most of the vacuum deposition techniques into the region
of the phase-diagram (Refs. [86] and [87]), where the material is tetragonal.
99
(The transition from tetragonal to orthorhombic is accomplished during a
subsequent anneal in oxygen at lower temperatures, typically around 450 °C to
500 °C and higher pressures, typically around several torr to atmospheric
pressure.) Learning more about the tetragonal phase by quenching the
material right after the deposition can tell us more about the processes during
the deposition and can help us improve the orthorhombic films.
6.2 Previous Work on Tetragonal YBa2Cu306.. Films
Because the tetragonal phase is not superconducting, it has met with
less interest than the superconducting phase. One of its applications is in
devices, where parts of a film are turned semiconducting in order to provide
an insulating area in the devices [88]. This is accomplished by laser-heating or
some other form of very localized heating. Li et al. [89] investigated the
influence of laser writing power densities and scanning speed as well as
different atmospheres (N2 or 02) on slightly oxygendeficient films (x --- 6.87)
and found that the oxygen content increased or decreased, depending on how
high a temperature the film reached during the annealing and how fast it
cooled through the optimum annealing temperature region of 400 °C to
500 °C.
As an interesting aside, some researchers also have sputterdeposited a
superconducting (5 = 0) tetragonal phase of YBa2Cu307_8 [90]. They claim that
due to the presence of activated oxygen species, the oxygen content of their
newly deposited films is higher than generally found. The tetragonal films
are quenched from the deposition temperature of 750 °C at a rate of
220 °C/min at a pressure of 1 bar [91]. These superconducting tetragonal
100
YBa2Cu307 thin films are very disordered, especially with respect to the
cations.
In addition, work has been done on the formation of secondary phases
and the decomposition of YBa2Cu306, This is important, because some of the
deposition techniques use conditions which are in a region of the phasediagram where YBa2Cu306, is unstable [15]. It is therefore important whether
and under what circumstances, any secondary phases form and if they do,
whether they are detectable in the orthorhombic phase and by what
techniques. Eibl and Roas [92] produced laser ablated films at reduced oxygen
pressures (10' mbar) and found BaCu2O2, Y203, Cu2Ox and Y2BaCu05, in TEM.
Interestingly, they never found Y2Cu2O5 under these conditions. Secondary
phases and impurities are important, because they might provide pinning
sites, therefore improving characteristics like J as pointed out by Tinkham
[93], on the other hand, disorder is detrimental to the Tc. The mechanisms of
formation have to be properly understood to tailor the material to ones
needs.
6.3 Starting Point for the Investigation
The starting point for the investigations in this thesis was some very
interesting data collected by Dr. Dennis Tom and Dr. Roland Platzer [27]. Their
PAC spectra on tetragonal YBa2Cu306.25 contained the signatures of the 'In
atoms experiencing two different welldefined electric field gradients, also
called sites. I will refer to the site already identified as the signal from the
YBa2Cu306., material as the A site and this new site as the B site. The B site is
remarkable for several reasons.
101
1) It was never seen in a spectrum of an orthorhombic film, regardless
of whether that film had been grown orthorhombic or whether it
had been grown tetragonal (and had shown the second site), and
annealed exsitu in oxygen to go through the phase transition.
Tetragonal films showed varying degrees of the B site: its appearance
was clearly very sensitive to the exact growth and insitu anneal
conditions.
2) Because of the tetragonal symmetry of YBa2Cu306.x, the material is
expected to have an asymmetry of ri = 0. Yet the A site has a
surprisingly high asymmetry of 1-1 = 0.4, but the second site has the
correct asymmetry of 11 = 0.
3) The main axis of the field gradient of the B site did not point along
any of the major axes of the YBa2Cu306,x, but was at an angle of 22°
with the c-axis of the YBa2Cu306,.x
Fig. 4.7 shows a typical orthorhombic spectrum where only one site is
seen. The same figure also shows a tetragonal spectrum without the B site.
Fig. 6.1 shows two tetragonal spectrum where the B site can be very clearly
seen. These three tetragonal spectra point to strong dependence of the
vo [MHz]
A site (YBa2Cu306.25)
B site
20(1)
133(1)
Table 6.5 Frequency and asymmetry of the A and B sites.
T1
0.40(5)
0
102
T4=500°C (lh at 10-6mbar)
6
5
4
3
2
AAA- 0
-6
1;=550°C (1h at 10 mbar)
-
6
5
4
3
2
1/4"-sZr/'. usunAni-VAn A
100
200
300
time [ns]
4000
100
200
300
400
500
[Mrad/s]
Fig. 6.1 R(t) and Fourier transform of two tetragonal films annealed at 500 °C
and 550 °C. These two spectra clearly show the influence of the annealing
conditions on the amount of B site formed.
amount of B site formed on the deposition and anneal conditions, something
which is not yet well understood.
At this time, the origin of the B site was unknown. One possible
explanation was that it could be an impurity formed during the deposition or
right afterwards, when the oxygen was removed and the films were cooled
down in vacuum. Another explanation, which held a lot of appeal, was that
the B site was the signal of YBa2Cu3O6, with exactly 6 oxygen atoms per unit
cell. Raman studies of the films had shown that their oxygen content was
approximately 6.25 and that the length of the anneal did not make
a
difference in the oxygen content. It was speculated that nanometer-sized (at
least) regions of YBa2Cu3O6 existed within YBa2Cu306.25 and that the PAC
103
technique might have detected this separation. (Other techniques, less local,
would not have detected this.) This explanation is consistent with the
observed asymmetry parameter, because the YBa2Cu306 phase is expected to
have B = 0, with no oxygen atoms in the basal plane and therefore n o
preferred direction, and is also consistent with the disappearance of the B site
on oxygenation, but it ultimately proved incorrect as will be seen later.
6.4 X-ray Studies on Tetragonal Films
In order to solve the second site mystery, a very careful x-ray study of
the tetragonal films was undertaken. Films which had been annealed in
vacuum for different amounts of time as well as quenched right after the
deposition or ramped down from the deposition temperature at a rate of
10°/min were used.
Careful investigation of the spectra showed two sets of very small
impurity peaks in the vicinity of the (004) and the (008) peaks, which
themselves are very small peaks in the x-ray spectrum of YBa2Cu307_5. Fig. 6.2
shows examples of the additional peaks for different kind of anneals, all of
them at a pressure of about 10' torn Except for the fact that the additional
peaks are in each of the spectra, there does not seem to be a particular pattern
to their peak height. This may mean that their formation is very sensitive to
the exact substrate temperature and oxygen pressure during deposition, or
some other possible factors like the stability of the deposition rate or the
quality of substrate surface.
104
8000
7000
6000
5000
co
4000
0°
3000
2000
1000
0
27
28
29
30
2
31
32
33
[O]
annealed for 3 h at 10-' torr (3h)
quenched from Ts (QCH)
annealed for 5 h at 1C torr (5h)
ramped from Ts (RMP)
Fig. 6.2 X-ray spectra of YBa2Cu306.25 films showing the two impurity peaks
near the (004) of YBa2Cu306.25. The ramp rate for the "ramped" film was
10 °C/min, the two annealed films were annealed at 500 °C for the times
indicated. All anneals were in a vacuum of 10-7 torn The spectra are not
normalized, but all have about the same peak-height for the YBa2Cu306.25
(004) peak.
Identification of secondary phases in an x-ray spectra is an art-form in
itself. In an oriented film, most of the reflections are missing and the
reflections that are there are not necessarily in the ratios given in the tables, if
they belong to different sets of planes. The identification of secondary phases
in YBa2Cu306, is further complicated by the fact that there are 4 elements,
which can form a large number of possible secondary phases. Table 6.6 gives
the 29 values and the accompanying d values for the impurity peaks observed
in several different samples.
105
6.5 Identification of the Second Site
For the identification, it was assumed that the two sets of peaks belong
to different orders of reflection from the same set of planes. This helped to
eliminate most of the possible candidates from a search performed on the set
of peaks around the (004) peak and the (008) peak separately and in the end
left only two possibilities, CuY0, in the (012) and (024) orientation,
corresponding to peak #2 and #4 in Table 6.6 and BaCup, in the (112) and
(224) orientation, corresponding to peak #1 and #3 in Table 6.6 (JCPDS files
#39-244 [94] and #39-245 [95]).
The identification of two impurities in x-ray diffraction must be
reconciled with the appearance of only one impurity in the PAC spectra. In all
the research done in Dr. Gardner's group [62], [55] it was shown repeatedly
that '11In substitutes readily for Y, preferring Y-rich phases. One of the possible
explanations is that both the indium and the yttrium have the same valence
3+ and they are also very close in ionic radius. Our x-ray investigation now
found two impurities, one containing yttrium, the second containing no
yttrium at all. We speculate that the In does not substitute in BaCu2O2 at all,
RMP
QCH
3h
5h
#1 [1 #1 [A]
#2 [0]
28.22
28.24
28.24
28.26
31.07
31.05
31.07
31.10
3.159
3.156
3.156
3.154
#2 [A] #3 [1 #3 [A]
2.875 _61.11 1.579
2.877
61.04 1.578
2.875
61.04 1.578
2.872
61.22 1.578
#4 [O]
64.83
62.50
62.55
62.53
#4 [A]
1.436
1.437
1.437
1.436
Table 6.6 20 values (in °) and the calculated d-values (in A) of the impurity
peaks for tetragonal YBa2Cu306.25 films subjected to different post-deposition
in-situ anneals. The names are explained in Fig. 6.2. Ref. [94] gives d=2.873 A
for the (012) peak of CuYO2 and 1.436 A for the (024), Ref. [95] finds the (112)
peak of BaCu2O2 at 3.155 A and the (224) peak at 1.578 A.
106
or that it substitutes at all of the many sites in BaCu2O2, thus diluting the
signal from any one site. Because PAC needs about 5% of the probe to be in
one site to identify the signal, if the indium substitutes in all possible sites, it
might not get over the detection limit for any single site and therefore be part
of the noise.
After having identified CuYO2 as a possible candidate for the B site,
does our PAC data agree with what is known about CuYO2? According to
Attili et al. [60] it does so very well. They found a vQ of 136 MHz, which agrees
with our measured data of 133 MHz and which makes the identification more
plausible.
As an aside, there are two different sets of data for the x-ray spectra of
CuYO2. As I have just shown, we can get an excellent agreement with one set
of published data [94], especially with respect to the orientation and confirm it
with PAC, but the second set of published data [96] does not give as good an
agreement in the 20 values and it assigns them to different h, k, 1 values.
Interestingly, this paper is used by Marshall et al. [97] to identify CuYO2 in
their films deposited at low oxygen pressures with a different orientation
than in ours.
Furthermore, because CuYO2 is rhombohedral, the field-gradient can be
assumed to point along the c-axis, with the same reasoning as in the case of
the tetragonal YBa2Cu3O6, because there is nothing to distinguish the al and
a2axis, but the c-axis is different from these axes. With this assumption, the
measured orientation of the field gradient (an angle of 22 ° of the main axis of
the electric field gradient with the c-axis of the YBa2Cu306.23) agrees with the
orientation seen in the x-ray spectra.
Additional evidence that the B site signal is from CuYO2 is that if a B
site signal is observed in the PAC spectrum, CuYO2 is observed in the 0-20
107
scan. As already mentioned, the B site signal was never observed in
orthorhombic films and likewise was CuYO2 never seen in 8 -28 spectra taken
on orthorhombic films. Furthermore, one tetragonal film with a B site was
annealed in flowing argon and up to 500 °C, the fraction of the B site
remained constant. At 600 °C and higher temperatures the B site disappeared
from the spectrum. After cooling the film in flowing argon and taken a 0-20
spectrum of it, no CuYO2 peaks were found in the spectrum.
Pressuretemperature phase diagrams are widely used to ascertain
which impurity phases are likely to appear in a material, but for the most part
are done on bulk samples and in air at temperatures higher (> 800 °C) than
what is used in our deposition system (< 700 °C). These phase diagrams are
used by researchers trying to grow crystals from a melt and are therefore
interested in the liquid phases. Yet it is surprising how little is known about
the phase equilibrium at reduced oxygen pressures and lower temperatures,
characteristic of typical thin film deposition conditions. Ahn et al.
[98]
investigated the phase diagram of YBa2Cu306. at 850 °C and reduced
pressures and found that CuO and BaCuO2 are no longer stable in contact
with YBa2Cu306,, but that BaCu2O2 becomes stable in contact with
YBa2Cu306, at low oxygen pressure, appearing first at a partial pressure of
oxygen of 3.0x10' atm
20 torr). BaCu2O2 does not exist along the BaO CuO
pseudobinary at temperatures between 900 °C and 1000 °C in air [99]. Based on
the appearance of different oxides at reduced oxygen pressures for the Cu Ba
O system, Beyers and Ahn. [100] suggest either a YBa2Cu306,--CuYO2 tie-line at
low pressures and temperature or they suggest that in offstoichiometry films
CuYO2 nucleates and grows faster than the Y2BaCuO5 and Cu20 expected from
the studies done in air at high temperatures. The faster nucleation
is
supported by the fact that the CuYO2 grows highly oriented, reported by
108
Marshall et al. [97] as well as in this thesis, even if the orientations do not
agree. Pankajavalli and Sreedharan [101] investigated the high temperature
stability of CuYO2 with oxide electrolyte e.m.f. measurements and found it
marginally stable with respect to dissociation into constituent binary oxides
over the range from room temperature to 1400 K. Konetzki and Schmid
Fetzer [102], using oxygen coulometric titration measurements in the
YCuO
found CuYO2 to be metastable and forming rather easily if the partial pressure
of oxygen is reduced. They almost sound annoyed at it, because it forms rather
easily, but the decomposition into the binary oxides is very sluggish, meaning
once it formed, it stayed around, even though it is only metastable. As a last
point, the Cu in both BaCu202 and CuYO2 is monovalent, which makes sense
under the reducing conditions of a very "reduced" oxygen pressure of less
than 10' torr.
All the above confirms that CuYO2 can and should form under
conditions of reduced oxygen pressures and high temperatures. Further
evidence for this was provided by the PAC spectra on the films with reduced
oxygen content, where in some of them the signal of CuYO2 appeared, even
though the annealing conditions were by far not as extreme as for the
tetragonal films (temperatures not above 600 °C and pressures greater than 10­
' torr (at 400 °C), at 600 °C, the pressures were even higher).
6.6 OxygenDeficient YBa2Cu306+X : Examples from Literature
As already pointed out in chapter 1, the oxygen content plays a crucial
role with respect to the characteristics of YBa2Cu306, The introductory part in
Chapter 1 was concerned with the differences between crystals and films,
109
especially pointing out that results cannot be easily carried over because of the
inherent differences in order between crystals and films. In this chapter I look
at some of the properties of oxygendeficient YBa2Cu306,. Chapter
1
illustrated the influence of the oxygen content on the T, and the c-axis length.
Several other properties have been investigated with respect to the oxygen
content.
Cochrane et al. [103] measured the Seebeck coefficient in oxygen
deficient polycrystalline YBa2Cu306. samples and found that for oxygen
contents 7-8 < 6.95 the carriers are positive (holes) but for very high oxygen
contents 7-8 > 6.95 the carriers are negative (electrons).
Ludwig et al. [104] found an increase in the London penetration depth
k from 175 nm for a YBa2Cu306.95 film to 380 nm for a YBa2Cu306.55 film. They
also measured the plasma frequency and found it to decrease from 9100 cm'
for the YBa2Cu306.95 film to 4200 cm-1 for the YBa2C U306.55 film.
Fietz et al. [105] investigated the influence of pressure on the T,. They
investigated the change in T, for polycrystalline and single crystal samples
stored at room temperature and for samples stored at 100 K. For both
temperatures they found a peak in dT,/dp at an oxygen content of 6.7. In
addition the samples stored at 298 K, with the lowest oxygen content being 6.4,
show an decreasing dT,/dp with increasing oxygen content. They explain
their results with two different effects; one is oxygen ordering within the
whole structure, facilitated by the increased pressure; the other is changing
distances between the oxygen chains and planes due to the applied pressure.
Because essentially no oxygen motion occurs at 100 K, they use their results at
this temperature to differentiate between the two different effects for the
298 K results.
110
6.7 Early Results: Anneals in Evaporator
To control the oxygen content in our films, two approaches were tried.
The first series of experiments was done in the deposition chamber itself,
using two different methods. For one set, after the deposition the pressure in
the chamber was brought up to the desired values (pressures of 1.5 torr,
1.5x10-1 torr, 7.7x10-1 torr, and 1.5x10-2 torr were used), and then the
temperature was lowered to the anneal temperature of 450 °C for 1 hour. Fig.
6.3 shows how the Tc and ATc change with decreasing anneal pressure. The
film annealed with the chamber backfilled to a pressure of 1.5x10' ton did not
have a transition at all, but showed semiconducting behavior.
During the anneal, the chamber pressure dropped significantly from
the original pressure (for the lower pressures by more than an order of
90
14
85
12
80
10
75
8
70
6
65
4
60
2
DT
55
0
10'2
10-1
10°
101
pressure[torr]
Fig. 6.3 Development of the Tc and ATc for films annealed at different
pressures, if the pressure is adjusted only once. The pressure was adjusted
while the films were still at the deposition temperature, then the films were
ramped down to the anneal temperature of 450 °C with no further
adjustment to the pressure. Films annealed at lower pressures had no
transition temperature.
111
magnitude). A second set of films was prepared with the pressure during the
initial cool down and anneal adjusted to stay at the desired value. For these
films, pressures of 15 torr, 1.5 torr, 1.5x10-1 torr, 1.5x10" torr, 7x10' torr and
1.5x10' torr were chosen. Fig. 6.4 shows how the T, and AT, change with
decreasing pressure. The transition width decreases by more than a factor of
two and the T, is much improved for the higher pressures. For the two lowest
anneal pressures, the films showed semiconducting behavior. (As an
interesting aside, the film annealed at 7x10' torr was measured right after the
deposition and showed a fairly high T, of 87.8 K and a AT of 3 K. Because it
did not look quite like a fully oxygenated film, but was speckled and brownish
instead of solidly black, the same film was remeasured a couple of days later
and then showed semiconducting behavior).
What can be learned from these two sets is that a surprisingly low
anneal pressure of 10' torr is enough to get a decent T,. Many published
90
6
88
5
4
86
T
3
C
OT
84
2
82
1
80
0.001
0
0.01
0.1
1
10
100
anneal pressure[torr]
Fig. 6.4 Development of T, and AT if the pressure is kept constant during the
cool-down and the anneal. For lower anneal pressures, the films did not have
a superconducting transition.
112
papers report that the chambers are backfilled to atmospheric pressure for the
post deposition anneal. In both sets, the films annealed at 1.5x104 torr still had
a T, of = 88 K. The most interesting region with T, = 60 K is not very easy to
reach and a T, of 60 K was reached only in the anneal which felt less
controlled. Furthermore, it was hard to control the pressure precisely, due to
the large volume of the chamber (84000 cm3) and the oxygen inlet and the
pressure gauge being far apart in the chamber. This was especially true for
anneal pressures between 10' ton (resulting in a film with a T, of 81.7 K) and
"3
torr (resulting in a film with semiconducting behavior), which can be
deduced as the region one should be able to reach all the intermediate T,
10
values. In addition, I was hesitant to use the cryo-pump to adjust the pressure
downward, for example to follow the proposed phase lines by Gallangher [28],
because either the pump might pump too much, but cannot be throttled in
the current setup, or a large influx of oxygen might overburden the pump.
Therefore a system, described in detail in section 2.5.1, was designed,
according to some suggestions in the literature [11], [12]. This system has a
much smaller volume of 4277 cm3 (about 5% of the evaporator volume).
Both references make a point about bulk material in the vicinity of the
sample. Ye and Nakamura [14] used it to prevent the film from becoming
contaminated with impurities in the gas flow, but also remarks that if they
tried to take out as much oxygen as possible, the films would decompose
without the bulk material. Co-evaporation is messy, because it evaporates the
metals not just on the substrate, but everywhere within the chamber, so some
bulk is close to the sample, but in a very uncontrolled way. In our new
system, a pellet of bulk YBa2Cu30, can be put right next to the film, giving one
more control about the amount and quality. Because of its much larger
volume, the pellet helps stabilize the oxygen content. Also taking the data
113
presented in Ref. [61], which shows significant barium loss at high
temperatures, the pellet with its much greater barium content might keep the
barium partial pressure high enough to keep the film from decomposing.
From data presented later in this chapter it can be inferred that whatever it
does, the pellet works better.
6.8 Results from the New System
The new system was very successful from the start and provided us
with means to reach Tc's down to 30 K, with values in between, exactly what
we had hoped to achieve.
The procedure is explained in more detail in section 2.5.2 and is based
on a phase diagram published in Ref. [28] and reproduced in Fig. 2.1, and gives
the pressuretemperature phase lines needed to achieve oxygen contents
between 6.2 and 6.7. A series of anneals was carried out, from a nominal
oxygen content of 6.4 to 6.8. All anneals included an annealing step at 600 °C,
except for the 6.8 film, which was done at 550 °C (doing it at higher
temperatures would have required overpressurizing the system). A second
annealing step was done at 450 °C, all the time staying on the phase line [28].
Afterwards, the samples were cooled down to 400 °C, still staying on the phase
line and then quenched by pulling out of the furnace. Due to some technical
difficulties with the cryogens, we were not able to see the resistive transition
in the 6.6 and 6.8 film, but in each case the onset of the transition was
observed and a rough guess for the Tc for the 6.6 film would be between 34 K
and 50 K and for the 6.8 film between 68 K and 75 K. The 6.7 film has a Tc of
52.45 K. What is really remarkable about the resistance curves is the fact that
they decrease almost linearly in resistance with a slope of almost 3 until they
114
140
1.2
120
1
100
0.8
80
60
0.6
40
0.4
20
0
0
50
100
150
T(K)
200
250
0.2
300
Fig. 6.5 Resistance of two films annealed on the 6.6 phase line, one with bulk
YBa2Cu3O7, the other with no bulk material in the vicinity. The structure in
the resistance of the "with pellet" film was repeatable during the
measurement itself, but might be an artifact of some measurement problems.
Due to contact and probe problems, the resistance measurement could not be
repeated after thermal cycling.
start going through the transition (Fig. 6.5), unlike the a-axis films where the
slope flattened, and sometimes the resistance started to rise again with
decreasing temperature before finally going through the transition.
To check for reproducibility, two 6.7 anneals on different films were
carried out. For these two films, the Tc's matched within 0.5 K and the change
in c-axis length matched within 0.01 A, which points towards excellent
reproducibility. Much to our surprise, this reproducibility was not evidenced
in the PAC films, a nominal 6.7 anneal gave a T, of 37.39 K for HTC 32 and of
47 K for HTC 34. But these films had suppressed T, after the deposition, and
therefore the widely different T, might be less a sign of nonreproducibility of
115
the annealing system and more an indication of problems during the
deposition.
Because the literature stresses the importance of YBa2Cu3O7 bulk in the
vicinity, one anneal (nominally 6.6) was done without any bulk. Even though
the anneal was otherwise identical to the same anneal with bulk material, the
electrical behavior was very different. The "bulk" film showed a transition,
while "no-bulk" film showed semiconducting behavior, clearly evident in
Fig. 6.5. At the same time, the c-axis length for the "no-bulk" 6.6 film is
11.86A, while the c-axis length for the "with bulk" film is 11.84 A, a difference
which is significant and which indicates together with the resistance behavior
that the oxygen content is lower than warranted by the anneal.
0.2
o
0.15
0
0
x
x
no bulk
0
O
0.1
0
x
0.05
o
o
0
0
bulk (1)
films (2)
films (3)
films: non PAC
films: PAC
00
x0 0 o
o0
0
I
6
6.2
6.4
6.6
6.8
oxygen content
Fig. 6.6 Change in c-axis length for PAC and non-PAC films and comparison
to published values. (1) is from Ref. [5], (2) is from Ref. [14], (3) is from Ref.
[12]. (1) and (2) match very well, as does our data with (3), but they do not
match with each other. The PAC and nonPAC match also, which is
important because the non PAC film underwent only one anneal, while the
PAC films were subjected to several.
116
The oxygen content of the films is not easily established. We have
access to two basic methods to determine the oxygen content: the c-axis
length, especially the change in the length, and for films above a certain
oxygen content, the transition temperature. The high temperature resistance
R(T)/R(300K) also appears to be systematically dependent on oxygen content,
but we have yet to investigate it fully.
Fig. 6.7 plots the transition temperature vs the nominal oxygen content
(as determined by the anneal conditions) and compares it to published data
[12], [14]. The Tc vs oxygen content curve published by Ye and Nakamura [14]
is very close to the curves published for bulk material and crystals, and i n
particular shows the 60K plateau so characteristic of bulk YBa2Cu306,, but
seldom seen in films. The curve of Ref. [12] has a very different behavior,
namely showing no evidence of a plateau. Our data falls in between these two
curves. Osquiguil et al. [12] in their paper are very careful about calling their
100
80
o
Osquiguil
o
Ye
0
non PAC films
oI
PAC films
60 -
0
0
0
0
0
40
0
0
20 -
no bulk
0
0
6.2
1
6.4
t
6.6
6.8
7
anneal
Fig. 6.7 Tc for our oxygendeficient YBa2Cu306, films compared to published
data. Our films fall between the two curves.
117
oxygen content "nominal", and the suppressed T, could indicate that the true
oxygen content might be lower than that suggested by the anneal.
More evidence for this is shown in Fig. 6.6. This graph plots the change
in c-axis length vs the oxygen content for powders [5] and films [12], [14]. The
change in c-axis length (as opposed to the absolute c-axis length) was chosen
to account for the generally elongated c-axis length in films, which can be
attributed to several factors, including cation disorder. Because it is unlikely
that our anneals at 600 °C changed the amount of cation disorder [106], the
contribution of the cation disorder to the elongation of the c-axis length
should be the same at any oxygen content.
Two important points are to be made about Fig. 6.6. Even though the
films made by Ye [14] seem to be very well equilibrated, judging from the T, vs
oxygen content curve, the change in c-axis length is slightly larger for films
than for the crystal data, again a sign that films are more disordered and
strained in general. The more important point is that while our data matches
very well for the oxygen content of 6.25 and for the 6.8 and higher, for the
range between 6.4 and 6.7 the curves do not match at all, but the change is
always larger, indicating again less oxygen in the material than what should
be in there according to the anneal. This deviation seems to depend on the
anneal itself and not the processing history, because the non-PAC films
usually were subjected to only one anneal, while the PAC films were
annealed several times without reoxygenation in between the different
anneals. Interestingly, the data taken by Osquiguil et al. [12] matches our data
fairly well, except perhaps for the 6.6 point, but even at this point their data is
clearly deviating from the published data by Jorgensen et al. [5] and Ye and
Nakamura [14]. At this point there is no good explanation for the differences,
especially because the anneal we used was closer to the method used by Ye
118
than Osquiguil, who surrounded the film completely with bulk material,
while Ye and we only had the bulk material in the vicinity. Osquiguil
considered the possible explanation of no oxygen ordering for their
suppressed Tc's unlikely (similar data had been observed in bulk materials
quenched from high temperatures [13]), because of the relatively low
quenching temperature. A second explanation was that the preparation time
was too short with respect to the oxygen diffusion coefficient and the films
did not have time to order properly. This is unlikely for the data presented by
Ye, who took about the same time to prepare their films. On the other hand,
in a paper published a year later by the group led by Bruynserade (which
includes Osquiguil) [107], they observe room temperature annealing for
oxygendeficient films. The change in T, is most striking for a film with a
nominal oxygen content of 6.55 which changes by 20 K. For oxygen contents
higher than 6.6, the T, changes by less than 4 K. The saturation T, is plotted vs
the oxygen content and a "guide to the eye" line is indicated, which shows the
2plateau behavior, but the spread in data is considerable. Room temperature
annealing was not investigated in our data, and the only real data to
investigate is in the case of a nominal oxygen content of 6.7. While the
non
PAC films usually were measured several days after the anneal, the PAC
films were measured no later than 24 hours after the anneal, but at this time
the films had almost reached their saturation T, according to [107]. The
differences in the T, between the PAC and non-PAC films are more likely due
to defects in the PAC films (they had suppressed T, right after the deposition)
than to room temperature annealing. Some of the non-PAC films are being
investigated by Raman-spectroscopy at the moment, which should tell us
more about the true oxygen content of the films. Whatever the physical
conditions that caused the differences to the data published by Ye, it happened
119
in two very different systems with film deposited by two different methods
(the Osquiguil films are sputtered, our films are co-evaporated).
6.9 PAC Results
The first set of PAC experiments on deoxygenated films at room
temperature showed that there is a significant change in frequency of the A
site with changing oxygen content as well as change in the orientation of the
electric field gradient. The latter is expected, because for fully oxygenated
films, the main axis of the electric field gradient is along the b-axis, while for
YBa2Cu306.25 it points along the c-axis. The question is, for what oxygen
Film Name
HTC 32
HTC 32
HTC 32
HTC 32
HTC 33
HTC 33
HTC 33
HTC 33
HTC 34
HTC 34
HTC 34
HTC 34
HTC 34
HTC 35
HTC 35
HTC 35
HTC 35
HTC 35
anneal
7 (as deposited)
6.7
6.5
6.4
7 (as deposited)
6.7
6.4
6.25
7 (as deposited)
6.7
6.5
6.4
6.25
7 (as deposited)
Tc [K]
79.9 K
37.9 K
n/a
n/a
82.55K
not measured
n/a
n/a
85.84
47
n/a
n/a
n/a
82.92
c-axis length [A]
11.7
11.77
11.835
11.841
not measured
not measured
not measured
not measured
11.705
11.782
11.837
11.849
11.859
11.719
6.7 @400 °C
6.7 @550 °C
6.7 @600 °C
6.7 @750 °C
Table 6.7 Summary of anneals performed on several PACfilms, as well as Tc
and c-axis length, when applicable.
120
content does it flip? We found that the main axis of the electric field gradient
had already flipped at a nominal oxygen content of 6.7 which was surprising,
but considering that the "6.7" is more likely to be a film with an oxygen
content of x = 6.5 considering the discussion earlier in this chapter, perhaps it
is not quite so unexpected. Because of inferior quality substrates, only some
preliminary information could be deduced, including the need to measure
the films in both orientations to be able to fit an ri to the data (The main axis
of the electric field gradient was pointing along the c-axis after the 6.7 anneal.
Fig. 4.6 shows clearly, that in that case for a = 45°, for 13 = 90° °2 is almost
completely suppressed and for 13 = 0°, a is almost completely suppressed. To
determine
from the ratio col /w2, both orientation have to be measured to
determine the frequencies.) Table 6.7 sums up the anneals done to the
different PAC films and gives the Ta's and measured c-axis length.
HTC 34 provided us with a lot of valuable data. As seen in the Table
6.7, we were able to perform a series of anneals, at a nominal oxygen contents
of x=7, 6.7, 6.5, 6.4 and 6.25. The 6.25 anneal was done by putting the film into
the deposition chamber, pumping down to a pressure of 10' torr, heating the
substrate to 500 °C for 1 hour and then quenching the film. Table 6.8 gives the
dependence of the asymmetry and the line width as well as the frequency on
the oxygen content.
121
The most important result is for a nominal oxygen content of 6.7,
which is more likely to be closer to 6.5, (further Raman studies are underway)
the asymmetry is 1, indicating that one of the components of the electric field
gradient is zero. Because the main axis of the electric field gradient points
along the c-axis, the zero component is in the ab-plane. Assuming an oxygen
content of 6.5 instead of the nominal 6.7, this would indicate a complete
ordering of the oxygen atoms into long chains, with every other chain
unfilled, as indicated in Fig. 1.6. For decreasing oxygen contents, the
asymmetry decreases again, indicating that the V. and Vyy become more
equal, which can be explained by oxygen moving out of the long ordered
chains. The line width is a measure of how welldefined the environment is
around the set of probes. The seeming contradiction between the asymmetry
indicating completely ordered chains and the line width widening at the
same time (see Fig. 6.9) can be resolved by considering that each probe which
contributes to the signal is in a site with an asymmetry of 1, which in turn
ends up forming the long chains, but because of small variations in the lattice
(the general theme of films being more disordered) the environments might
be slightly different, leading to a bigger linewidth. Because there are obvious
changes within the lattice (the c-axis length increases with decreasing oxygen
oxygen content
7
6.7
6.5
6.4
6.25
vQ [MHz]
36.0(1)
24.0(1)
24.3(1)
21.5(1)
19.6(1)
T1
0.51(1)
0.99(1)
0.50(1)
0.46(1)
0.34(1)
Aco/co [ %]
12.2
19.4
21.5
12.8
12.8
Table 6.8 vQ, rl, Ace/co for film HTC 34 annealed for nominal oxygen contents
of 7 (as deposited), 6.7, 6.5, 6.4 and 6.25.
122
-0.15
I
1
HTC34 2000A thin film
-0.10
F
I
-
07-45, ft=90 YBaCuO7
= 0,13 =90
TeRT
.
« =45 1;=0 vo=38.0(1)MHz
-0.05
a=0, 13 :-.0
.1%
0.00
4.4
i=.51(1)
.6o/co=12.2%
/44W
e
2
)
-0.15
0
-0.10 ;
-6
YBaCu0e7
va=24.0(1)MHz
..
-0.05
-
0.00
\
Eu
of
YBa Cu0,5
va=24.3(1)MHz
rr--.50(1)
-0.10
6
4
aok.o=21.5%
2
0.00
-0.15
,C.y
...,,,,
,/,', i
I
-
1
0
-6
YBaCu0,,,
-0.10
va=21.5(1)MHz
i=.48(1)
-0.05
4
2
-0.15
-0.05 r
13
-
_
11=99(1)
boico=19.4%
Navoii1/40,446041#4
.6co/o3=12.8%
0.00
"-;
-0.15
I
- 4
2
f
I
1
-1­
`-l-- 1 '.,I­1­
YBaCuO,
va=19.8(1)MHz
Tr .34(1)
-0.10
V0
-6
_
Ar.o/o3=12.8%
4
2
0.00
0
100
200
time [ns]
300
0
50
100
150
200
2
[Mrad/s]
Fig. 6.8 R (t) curves and Fourier transforms of all of the spectra taken on HTC
34, with the anneals indicated. All spectra were taken at room temperature.
content), this could explain why the line width changes, but it would require
more investigation.
123
A final set of experiments was done by moving the annealing set up to
the spectrometers and, using a different furnace, observing the PAC spectra
insitu while keeping the films at the various temperatures and appropriate
pressures to reach a nominal 6.7 oxygen content. It was hoped to see some
dynamic effects, particularly seeing the oxygen move, which allows the
calculation of hopping rates. The oxygen motion influences the linewidth,
and the line first widens with increasing temperature due to the thermal
motion. At a certain temperature, the motion becomes faster than the
lifetime of the probe nucleus and the line width narrows again. This
phenomenon is called motional narrowing. Wang [108] in her thesis
measured this behavior in ceria and has an overview of the literature.
The motional narrowing was not observed in HTC 35, the linewidth
stayed the same over the whole range from 400 °C to 750 °C , as can be seen in
Fig. 6.10. The reason for this behavior is not clear at the moment, but one can
1.1
24
1
22
0.9
20
0.8
11
18
Awko
0.7
16
0.6
14
0.5
0.4
12
0.3
10
6.2
6.4
6.6
6.8
7
oxygen content
Fig. 6.9 Asymmetry and linewidth as a function of oxygen content for HTC 34.
While the asymmetry has a maximum for an oxygen content of 6.7, the
relative linewidth is the biggest for an oxygen content of 6.5.
124
speculate. One reason might be that we had already reached the tailend of
the motional narrowing and that the linewidth could not narrow any more.
Furthermore, the above behavior was calculated for materials with the
symmetry axis of the electric field gradient fluctuating randomly in all three
dimensions [109] but in the YBa2Cu30,_5 we have the oxygen moving in only
two dimensions. These guesses show that more investigation has to be done,
including giving the theory a chance to catch up and calculate what one
would expect to see.
We learned something else from this film, because in our bid to see
some dynamic effects, we heated the sample up to 750 °C (while slightly
overpressurizing it) and did not see any decomposition, in contrast to early
films where decomposition began at 700 °C, when heated in flowing oxygen
(at 800 °C, the amount of Y2Cu2O5 was significant). This indicates that the
YBa2Cu306, pellet kept the sample from decomposing, most probably by
providing a high enough barium partial pressure to keep the barium in the
film (the vapor pressure of barium at .--775 °C is of the order of 10-1 torr [110]).
These experiments on oxygendeficient YBa2Cu306, films showed
some very promising first results. The most important step now is to
determine the oxygen content of the films by Raman spectroscopy. In
addition, the oxygen contents between the nominal 6.7 and the fully
oxygenated 7.0 need to be investigated, especially with respect to the
asymmetry. Our last set of experiments also indicates that oxygendeficient
films should be investigated below 400 °C, because some significant oxygen
motion might be occurring at temperatures below 400 °C. Our films were
quenched from 400 °C, assuming that the oxygen atoms were moving too
slow at lower temperatures to cause significant uptake during cooldown and
125
therefore increasing the oxygen content, which they did not do judging from
the measured T, and c-axis length.
-0.15
-0.10
-0.05
0.00
HTC35 2000A thin film
I
I
v0=37.2(1 )MHz
n=.44(1)
: .o.owlwxy. .
*1.,.."'
I
I
I
AcuAo=13.1%
,,-;
I
TM 400°C
-0.10
LL
0.00
-0.15
YBaCuO
t
a=0, 13=0
...
_
2
-a
0
6
4
15%
Neell0WW:44L0,641
4,;
i
f
1
t
I
2
f
0
6
I
Tm=600°C
-0.10
4
-0.05
1!
2
:.1.,
0.00
-0.15
0
6
_
0.00
-0.15
2
<$.45, li=0
Tm=550°C
-0.10
4
4 --%
lilt!
-0.05
6
ToRT
.:\
Negotovifi4"4440400.Olt
-0.15
-0.05
YBaCuO7
1
i
0
6
,
1
I
Tm=650°C
-0.10
-0.05
'
A.t...1
0.00
.
,
s
,
,
,
4
;
I ',4
1
I,
4
_...
2
6
I
I
100
200
time [ns]
.
1
300
50
100
03
150
200
0
250
[Mrad/s]
Fig. 6.10 R (t) and Fourier transforms for HTC 35, annealed insitu at different
temperatures to achieve a 6.7 anneal. The pressure was kept at the value
appropriate according to the phasediagramm [281
126
Chapter 7
Summary
7.1 The Electric Field Gradient in YBa2Cu307_8 Films
Perturbed angular correlation spectroscopy (PAC) is used as a tool to
study phase transition in materials, because its sensitivity is temperature
independent, and it can therefore be used to very high temperatures (1000 °C),
unlike some of the related hyperfine techniques of nuclear quadrupole
resonance (NQR) and MoBbauer spectroscopy. For the study of thin films,
PAC has the added advantage that the number of probe atoms needed (1010 to
1011) to perform the measurements is several orders of magnitudes smaller
than for MoBbauer spectroscopy (10') and NQR (1018).
All of these hyperfine techniques measure the interaction of the
quadrupole moment of the probe nucleus with electric field gradient at the
site of the nucleus. The electric field gradient stems from the distribution of
charges around the probe nucleus and is a tensor. It characterizes the
asphericity of the charge distribution. As a consequence, in cubic materials the
electric field gradient is zero, and hyperfine techniques can be used in cubic
materials to study defects.
It is possible to do PAC studies in powders as well as oriented samples.
If oriented samples are used, either in the form of single crystals or thin films,
the orientation of the electric field gradient with respect to the host crystal can
be determined. This thesis completed the determination of the orientation of
the electric field gradient, in orthorhombic YBa2Cu30, as well as tetragonal
YBa2Cu306.25 at room temperature and ambient pressure. Table 7.9 is a
summary of the information determined.
127
YBa2Cu3O7
YBa2Cu30625
main axis of efg
along b-axis
along c-axis
vo [MHz]
39
19
1
0.5
0.4
Table 7.9 Summary of electric field gradient in YBa2Cu3O7 and YBa2Cu306.25 at
room temperature.
c-Axis oriented YBa2Cu306, films are twinned. This required the
growth of a-axis oriented films, because the main axis of the electric field
gradient points along the b-axis. This conclusion could only be reached if
about 20% of the film were assumed to be b-axis oriented, an assumption
which could not be confirmed by x-ray diffraction measurements. This issue
has not been conclusively resolved, but arguments based on the differences
between x-ray diffraction and PAC can be used to explain it. For YBa2Cu306.25
films, the direction of the main axis of the electric field gradient is readily
determined from c-axis oriented films, because it is pointing along the c-axis.
To determine the optimum growth conditions for a-axis films, studies
on the dependence of the amount of a-axis oriented grains depending on the
temperature and pressure were carried out. It was shown that the largest
amount of a-axis oriented grains was formed at a deposition temperature of
600 °C and a local oxygen pressure at the substrate of 1.1 mtorr, but under
these conditions the Tc was suppressed to 58 K.
7.2 Impurity Phases in YBa2Cu306.x
Much of the early PAC work in YBa2Cu306x was plagued by uncertainty
about which is the true YBa2Cu306+), signal and at what site within the
structure does the 111In substitute. In retrospect, we see that much of the
128
earlier work determined signals from impurity phases and not from the
YBa2Cu306+x itself.
Once all of the PAC signals of possible impurity phases had been
identified, PAC is a useful tool to determine the quality of YBa2Cu306+x films
and can give early warning signals about pressure and temperature
conditions which lead to degradation of this films. PAC has advantages over
x-ray diffraction, a common tool to determine the existence of impurity
phases, because it is more local. In addition, x-ray diffraction can only show
the existence of oriented phases, but an impurity will show a PAC signal
regardless of orientation.
Orthorhombic YBa2Cu307_5 films deposited in this laboratory showed
only one site, which had been identified in earlier work as the signal of the
'In substituting at the Y site. None of the commonly seen impurities in
YBa2Cu307_,5 like Y2Cu2O5 and Y2BaCuO5 were seen in our PAC spectra.
Tetragonal YBa2Cu306.25 films, which were either quenched from the
deposition conditions or were subjected to an additional anneal at 10-7 torr
and 500 °C, showed a second site, called the B site. This signal was identified
as CuYO2 growing in the (012) orientation. Supporting x-ray studies confirmed
this and also showed the formation of BaCu2O2 in the (112) orientation.
The B site signal was never seen in orthorhombic films and it
disappeared in films when they were annealed in oxygen. It was also shown
to disappear when the film was annealed in argon around 600 °C.
Gaining as much information as possible about the formation of the B
site is important, because it forms under conditions commonly employed in
the deposition of YBa2Cu306, films, which are deposited in the tetragonal
phase and are transformed to the superconducting orthorhombic phase by
annealing them in oxygen at temperatures around 450 °C. Being able to prove
129
the formation of CuYO2 under conditions not far removed from the actual
deposition could mean, that even under standard condition CuYO2 forms, but
in amounts to small to detect.
7.3 Oxygen-deficient Films
YBa2Cu306+x
films with varying oxygen contents were made to
determine the change in the electric field gradient between a fully oxygenated
YBa2Cu3O7 and a YBa2Cu306.25 film. To change the oxygen content a system
was built which allows fine control of the pressure and the temperature. The
films were annealed according to a phase diagram proposed by Gallagher [28],
by staying on the pressure-temperature phase line for a given oxygen content.
The anneals are a two-step process, with one annealing step at 600 °C and a
second one at 450 °C. After the anneal the films were quenched from 400 °C to
room temperature.
We have not yet verified that the oxygen content indicated by the
anneal conditions was indeed obtained in our films. We suspect that the
films have lower oxygen content than that intended. This suspicion is based
on reduced Tc's compared to published data as well as the observed change in
the c-axis length. Comparing our data to published data, we find a
surprisingly good match to published film data [12], but both theirs and our
data do not match published data for crystals [5] for oxygen contents between
6.7 and 6.3.
We found that the frequency vQ decreased with decreasing oxygen
content. A surprising result was the behavior of the asymmetry. The
asymmetry is 1 at an nominal oxygen content of 6.7 and then decreases again.
X-ray as well as Tc measurement indicate that this oxygen content might be 6.5
130
(Raman studies are under way) and that the result can be explained by a
complete ordering of the oxygen atoms along the b-axis with every other row
unoccupied.
Insitu anneals, where the pressure and temperature were controlled
while the sample was in the spectrometer did not show the motional
narrowing expected, which would be evidenced in the line width.
7.4 Future Direction
In order to fully understand our results, the oxygen content of the films
has to be determined. X-ray studies as well as the measured Tc's (where
applicable) indicate that the actual oxygen contents are lower than they
should be according to the anneal. This might work to our advantage, because
there is every indication that the nominal 6.7 film is actually a 6.5 film, which
is an oxygen content we are very much interested in.
This thesis presented some very promising first results, but also shows
that there is a lot more to explore. The region between the nominal oxygen
content of 6.5 and the fully oxygenated YBa2Cu3O7 has to be carefully
investigated to determine the change in the asymmetry. From the behavior of
the asymmetry, one should be able to conclude more about the oxygen
ordering in YBa2Cu306,. In order to do this, some experimental difficulties
have to be solved for films with an oxygen content higher than 6.7, because
the current procedure would require me to overpressurize the system.
However the region with oxygen content less than 6.7 should be easily
accessible.
Our first try at dynamic studies showed that we looked in the wrong
place and did not see any dynamic behavior. This problem now has to be
131
attacked on two different fronts. One is the theoretical front, what we expected
to see was deduced from behavior seen in ceria. Because YBa2Cu306, is very
different from ceria, starting with the crystal structure, theoretical calculations
need to be carried out to calculate what is expected. On the experimental front,
the region between room-temperature and 400 °C has to be investigated.
Oxygen diffusion is very slow at these temperature, but for oxygen ordering to
occur, the oxygen atoms only have to move on the order of one unit cell,
which is definitely possible at room temperature, as evidenced by room
temperature annealing.
The experiments carried out in this thesis answered some questions,
like the final determination of the direction of the main axis of the electric
field gradient, but as it normally happens in science, they provided us with a
whole new set of questions to be answered, especially with respect to the work
on oxygendeficient films.
132
Bibliography
1.
Fiory, A.T., Martin, S., Fleming, R.M., Schneemeyer, L.F., Waszczak,
J.V., Hebard, A.F., and Sunshine, S.A., Physica C, (1989), 162-164, 1195.
2.
Onnes, K.H., Commun. Phys. Lab. Univ. Leiden, (1911) 124c.
3.
Bednorz, J.G. and Muller, K.A., Zeitschrift fur Physik B., (1986) 64, 189.
4.
Wu, M.K., Ashburn, J.R., Torng, C.J., Hor, P.M., Meng, R.L., Gao, L.,
Huang, Z.J., Wang, Y.Q., and Chu, C.W., Phys. Rev. Lett., (1987) 58(9),
908
5.
Jorgensen, J.D., Veal, B.W., Paulikas, A.P., Nowicki, L.J., Crabtree, G.W.,
Claus, H., and Kwok, W.K., Phys. Rev. B, (1990) 41(4), 1863.
6.
Poole, C.P.J., Farach, H.A., and Creswick, R.J., Superconductivity. (1995),
San Diego, Academic Press.
7.
Neiman, R.L., Giapintzakis, J., and Ginsberg, D.M., Phys. Rev. B, (1994)
50(21), 16028.
8.
Burmann., T., Geerk, J., Meyer, 0., Schneider, R., and Linker, G., Solid
State Communications (1994) 90(9), 599.
9.
Karppinen, M., Fukuoka, A., Niinisto, L., and Yamauchi, H.,
Superconductor Science and Technology, (1996) 9(3), 121.
10.
Appelman, E.H., Morss, L.R., Kini, A.M., Geiser, U., Umezawa, A.,
Crabtree, G.W., and Carlson, K.D., Inorg. Chem., (1987) 26(20), 3237.
11.
Nakamura, K., Ye, J., and Ishii, A., Physica C, (1991) 213, 1.
12.
Osquiguil, E., Maenhoudt, M., Wuyts, B., and Bruynseraede, Y., Appl.
Phys. Lett., (1992) 60(13), 1627.
13.
Namgung, C., Irvine, J.T.S., and West, A.R., Physica C, (1990) 168, 346.
14.
Ye, J. and Nakamura, K., Phys. Rev. B, (1993) 48(10), 7554.
15.
Matijasevic, V., Rosenthal, P., Shinohara, K., Marshall, A.F.,
Hammond, R.H., and Beasley, M.R., J. Mater. Res., (1991) 6(4), 682.
16.
Martinez, 0., Jimenez, J., Martin, P., Prieto, A.C., Degoy, S.,
Chambonnet, D., Belouet, C., Nicoletti, S., and Correra, L., Physica C,
(1996) 270, 144.
133
17.
Huong, P.V., Bruyere, J.C., Bustarret, E., and Granchamp, P., Solid State
Communications, (1989) 72(2), 191.
18.
Ceder, G., McCormack, R., and de Fontaine, D., Phys. Rev. B, (1991)
44(5), 2377.
19.
Cava, R.J., Batlogg, B., Rabe, K.M., Rietman, E.A., Gallagher, P.K., and
Rupp, L.W.J., Physica C, (1988) 156, 523.
20.
Liitgemeier, H., Schmenn, S., Meuffels, P., Storz, 0., Schollhorn, R.,
Niedermayer, C., Heinmaa, I., and Baikov, Y., Physica C, (1996) 267, 191.
21.
de Fontaine, D., Wille, L.T., and Moss, S.C., Phys. Rev. B, (1987) 36(10),
5709.
22.
,
Aligia, A.A., Garces, J., and Bonadeo, H., Phys. Rev. B, (1990) 42(16),
10226.
23.
Veal, B.W., Paulikas, A.P., You, H., Shi, H., Fang, Y., and Downey, J.W.,
Phys. Rev. B, (1990) 42(10), 6305.
24.
Shaked, H., Jorgensen, J.D., Hunter, B.A., Hitterman, R.L., Paulikas,
A.P., and Veal, B.W., Phys. Rev. B, (1995) 51(1), 547.
25.
Shin, D.C., Park, Y.K., Park, J.-C.L., Kang, S.-J.1., and Yonn, D.Y., Physica
C, (1997) 280, 142.
26.
Maissel, L.I. and Glang, R., eds. Handbook of Thin Film Technology.
(1970), McGraw-Hill Book Company, New York.
27.
Tom, D.W., Microstructural Characterization of YBa2Cu307_,5 Thin
Films with Time Differential Perturbed Angular Correlation
Spectroscopy, PhD thesis in Physics. (1995), Oregon State University,
Corvallis, Oregon
28.
Gallagher, P.K., Advanced Ceramic Materials, (1987) 2(3B), 632.
29.
Thomsen, C. and Cardona, M., Raman Scattering in High-Tc
Superconductors, in Physical Properties of High Temperature
Superconductors I, D.M. Ginsberg, Editor. (1989) World Scientific,
Singapore.
30.
Dieckmann, N., Kiirsten, R., Lohndorf, M., and Bock, A., Physcia C,
(1995). 245, 212.
31.
Feile, R., Physica C, (1989) 159, 1.
134
32.
Macfarlane, R.M., Rosen, H.J., Engler, E.M., Jacowitz, R.D., and Lee,
V.Y., Phys. Rev. B, (1988) 38(1), 284.
33.
Thomsen, C., Wegerer, R., Habermeier, H.-U., and Cardona, M., Solid
State Communications, (1992) 83(3), 199.
34.
Gibson, G., Mac Manus-Driscoll, J.L., and Cohen, L.F., IEEE Transactions
on Applied Superconductivity, (1997) 7(2), 2130.
35.
Cohen, L.F., Li, Y.B., Gibson, G., and Mac Manus-Driscoll, J.L.
Investigation of cation disorder in c-axis YBCO 123 thin films using
Raman miscroscopy, in MRS (Fall Meeting). (1995) Boston.
36.
Schatz, G. and Weidinger, A., Nuclear condensed matter physics:
nuclear methods and applications. 2 nd ed. (1996), Chichester; John
Wiley & Sons Ltd.
37.
Ito, W., Mahajan, S., Yoshida, Y., Morishita, T., Kumagai, M., and
Yabuta, K., Jpn. J. Appl. Phys., (1994) 33(Part 1), 5701.
38.
Ashcroft, N.W. and Mermin, N.D., Solid State Physic. 1976: Holt,
Rinehart and Winston, 826.
39.
Scherrer, P., Nachr. Ges. Wiss. Gottingen, Math.-Phys. Kl., (1918) 2, 96.
40.
Prakash, S. and Bunshah, R.F., J. Mater. Res., (1992) 7(9), 2343.
41.
Nuffield, E.W., X-ray Diffraction Methods. 1966: John Wiley & Sons,
Inc.
42.
Cullity, B.D., Elements of X-ray Diffraction. 2nd ed. 1978: AddisonWesley Publishing Company, Inc.
43.
Knorr, D.B., Weiland, H., and Szpunar, J.A., JOM, (1994,September), 32.
44.
Zheng, J.P., Dong, S.Y., and Kwok, H.S., Appl. Phys. Lett., (1991) 58(5),
540.
45.
Frauenfelder, H. and Steffen, R.M., in Alpha-, Beta- and Gamma-Ray
Spectroscopy, K. Siegbahn, Editor. (1965), North-Holland: Amsterdam.
p. 997.
46.
Butz, T., Hyperfine Interact., (1989) 52, 189.
47.
Wegner, D., Hyperfine Interact., 91985) 23, 179.
48.
Yu, J., Freeman, A.J., Podloucky, R., Herzig, P., and Weinberger, P.,
Phys. Rev. B, (1991) 43(1), 532.
135
49.
Arends, A.R., Hohenemser, C., Pleiter, F., De Waard, H., Chow, L., and
Suter, R.M., Hyperfine Interact., (1980) 8, 191.
50.
Masterov, V.F., Nasredinov, F.S., and Seregin, P.P., Phys. Solid State,
(1995) 37(5), 687.
51.
52.
53.
Trager, W. and Butz, T., Z. Naturforsch., (1992) 47a, 12.
Bartos, A. and Uhrmacher, M., Phys. Rev. B, (1993) 48(10), 7478.
Saitovitch, H., Silva, P.R.J., and Rodriguez, A.M., Hyperfine Interact.,
(1992) 73, 277.
54.
Schwenker, R., Perturbed angular correlation spectroscopy in the high
temperature superconducting material YBa2Cu307_x, MS thesis in
Physics. (1990), Oregon State University, Corvallis,Oregon.
55.
Fassel, A., Chemical Phases in the Y-Ba-Cu-O System Identified by
TDPAC, MS thesis in Physics. (1993), Oregon State University:
Corvallis, Oregon.
56.
Tom, D.W., Platzer, R., Gardner, J.A., and Tate, J., Appl. Phys. Lett.,
(1993) 63(23), 3224.
57.
Bartos, A. and Uhrmacher, M., Hyperfine Interact., (1990) 61, 1101.
58.
Uhrmacher, M. and Bartos, A., Hyperfine Interact., (1990) 61, 1097.
59.
Wiarda, D., Uhrmacher, M., Bartos, A., and Lieb, K.P., J. Phys.: Condens.
Matter, (1993) 5, 4111.
60.
Attili, R.N., Uhrmacher, M., Lieb, K.P., and Ziegeler, L., Phys. Rev. B,
1996. 53(2), 600.
61.
Platzer, R., Schwenker, R., Passel, A., Tom, D.W., Tate, J., Gardner, J.A.,
Evenson, W.E., and Sommers, J.A., Hyperfine Interact., (1997) 110, 271.
62.
Weidlich, G.A., Structural Properties of Indium Doped YBa2Cu307_x
High Tc Superconductors, MS thesis in Physics. 1989, Oregon State
University, Corvallis.
63.
Ambrosch- Draxl, C., Blaha, P., and Schwarz, K., Hyperfine Interact.,
(1990) 61, 1117.
64.
Plank, H., Bauer, 0., Forkel, D., Meyer, F., Roas, B., Saemann-Ischenko,
G., Strobel, J., Wolf, H., and Witthuhn, W., Hyperfine Interact., (1990)
61, 1139.
136
65.
Platzer, R., Dumkow, I.D., Tom, D.W., Tate, J., and Gardner, J.A., J.
Mater. Res., (1998), accepted for publication March 1998.
66.
Braginski, A.I., Physica C, (1991) 185-189(1), 391.
67.
Ito, W., Mahajan, S., Yoshida, Y., Watanabe, N., and Morishita, T., J.
Mater. Res., (1994) 9(7), 1625.
68.
Eom, C.B., Marshall, A.F., Laderman, S.S., Jacowitz, R.D., and Geballe,
T.H., Science, (1990) 249, 1549.
69.
Li, Y.Q., Zhao, J., Chern, C.S., Gal lois, B., Norris, P., and Kear, B., J.
Appl. Phys., (1992) 71(5), 2472.
70.
Jeschke, U., Schneider, R., Ulmer, G., and Linker, G., Physica C, (1995)
243, 243.
71.
Terashima, Y., Sagoi, M., Kubo, K., Mizutani, Y., Miura, T., Yoshida, J.,
and Mizushima, K., Jpn. J. Appl. Phys., (1989) 28(4), L653.
72.
Thong, Q., Chou, P.C., Li, Q.L., Taraldsen, G.S., and Ignatiev, A., Physica
C, (1995) 246, 288.
73.
Raven, M.S., Inameti, E.E., Wan, Y.M., and Murray, B.G., phys. stat. sol.
(a), (1994) 146, 685.
74.
Inam, A., Rogers, C.T., Ramesh, R., Remschnig, K., Farrow, L., Hart, D.,
Venkatesan, T., and Wilkens, B., Appl. Phys. Lett., (1990) 57(23), 2484.
75.
Trajanovic, Z., Takeuchi, I., Warburton, P.A., Lobb, C.J., and
Venkatesan, T., Physica C, (1996) 265, 79.
76.
Rosova, A., Chromik, S., Benacka, S., and Wuyts, B., Physica C, (1995)
253, 39.
77.
Sodtke, E. and Miinder, H., Appl. Phys. Lett., (1992) 60(13), 1630.
78.
Hamet, J.F., Mercey, B., Hervieu, M., and Raveau, B., Physica C, (1992)
193, 465.
79.
Ha, D.H., Park, Y.K., and Park, J.-C., Jpn. J. Appl. Phys., (1994) 33, L1588.
80.
Mahajan, S., Ito, W., Yoshida, Y., and Morishita, T., Physica C, (1993)
213, 445.
81.
Kawamoto, K. and Hirabayashi, I., Oxygen ordering and transformation
from a-axis oriented to b-axis oriented films in YBa2Cu30y., in
Advances in Superconductivity VII. 1994.
137
82.
83.
Ronay, M. and Nordlander, P., Physica C, (1988) 153-155, 834.
Mogro-Campero, A. and Turner, L.G., Journal of Superconductivity,
(1991) 4(4), 279.
84.
Filipczuk, S.W., Driver, R., and Smith, G.B., Physica C, (1990) 170, 457.
85.
Siegal, M.P., Philips, J.M., van Dover, R.B., Tiefel, T.H., and Marshall,
J.H., J. Appl. Phys., (1990) 68, 6353.
86.
Bormann, R. and Milting, J., Appl. Phys. Lett., (1989) 54, 2148.
87.
Hammond, R.H. and Bormann, R., Physica C, (1989) 162, 703.
88.
Yoshida, A., Tamura, H., Gotoh, K., Imamura, T., and Hasuo, S., J.
Appl. Phys., (1993) 73(11), 7549.
89.
Li, Y.B., Shelley, C., Cohen, L.F., Cap lin, A.D., and Stradling, R.A., J.
Appl. Phys., (1996) 80(5), 2929.
90.
Lopez, J.G., Siejka, J., Mercandalli, L.M., Bisaro, R., Blanchin, M.G., and
Teodorescu, V., Physica C, (1997) 275, 65.
91.
Blanchin, M.-G., Teodorescu, V.S., Garcia-Lopez, J., Siejka, J., Bisaro, R.,
and Mercandalli, L.M., Phil. Mag. A, (1996) 74(1), 151.
92.
Eibl, 0. and Roas, B., J. Mater. Res., (1990) 5(11), 2620.
93.
Tinkham, M., Hely. Phys. Acta, (1988) 61, 443.
94.
Kohler, B. and Jansen, M., Z. Anorg. Aug. Chem., (1986) 543, 73.
95.
Teske, C.L. and Miiller-Buschbaum, H., Z. Naturforsch., B: Anorg.
Chem., Org. Chem., (1972) 27, 296.
96.
Ishiguro, T., Ishizawa, N., Mizutani, N., and Kato, M., J. Solid State
Chem., (1983) 49, 232.
97.
Marshall, A.F., Matijasevic, V., Rosenthal, P., Shinohara, K.,
Hammond, R.H., and Beasley, M.R., Appl. Phys. Lett., (1990) 57(11),
1158.
98.
Ahn, B.T., Lee, V.Y., Beyers, R., Giir, T.M., and Huggins, R.A., Physica
C, (1990) 167, 529.
99.
Whitler, J.D. and Roth, R.S., Phase Diagrams for High Tc
superconductors, American Ceramic Society, Washington, (1991).
138
100.
Beyers, R. and Ahn, B.T., Annu. Rev. Mater. Sci., (1991) 21, 335.
101.
Pankajavalli, R. and Sreedharan, 0.M., Journal of Materials Science,
(1996) 31, 3137.
102.
Konetzki, R.A. and Schmid-Fetzer, R., J. Solid State Chem., (1995) 114,
420.
103.
Cochrane, J.W., Russell, G.J., and Matthews, D.N., Physica C, (1994) 232,
89.
104.
Ludwig, C., Jiang, Q., Kuhl, J., and Zegenhagen, J., Physica C, (1996) 269,
249.
105.
Fietz, W.H., Quenzel, R., Ludwig, H.A., Grube, K., Schlachter, S.I.,
Hornung, F.W., Wolf, T., Erb, A., Klaser, M., and
ller-Vogt, G.,
Physica C, (1996) 270, 258.
106.
Ye, J. and Nakamura, K., Physica C, (1995) 254, 113.
107.
Libbrecht, S., Osquiguil, E., Wuyts, B., Maenhoudt, M., Gao, Z.X., and
Bruynseraede, Y., Physica C, (1993) 206, 51.
108.
Wang, R., Perturbed Angular Correlation Spectroscopy of Defects in
Ceria, PhD thesis in Physics. 1991, Oregon State University, Corvallis
109.
Evenson, W.E., Mc Kale, A.G., Su, H.T., and Gardner, J.A., Hyperfine
Interact., (1990) 61, 1379.
110.
Lide, D.R., ed. CRC Handbook of Chemistry and Physics. 73rd ed. (1993),
CRC, Boca Raton.
139
Appendices
140
Appendix A
How to grow a film
The growth process of a films can be split up into several steps
choice and cleaning of the substrate
loading the chamber
heating the substrate in preparation for the actual film growth
depositing the film
annealing the film
cool-down after deposition
A.1 Choice and Cleaning of the Substrate
Three main substrates are being used in this lab to grow YBa2Cu306,
films on them. Magnesiumoxide (MgO) can be considered the workhorse
substrate. It is used especially when a new procedure is being tried and one
goes through a lot of substrates trying to determine the optimum parameters.
MgO is significantly cheaper than the other substrates. Its disadvantages are
that it has a fairly large lattice mismatch (a=4.213 A, which is a mismatch of
10.2% with respect to the a-axis and 8.2% with respect to the b-axis) and it is
hydroscopic and has a tendency to form an amorphous surface layer after a
fairly short time. One can still grow films on these substrates, but their quality
is usually not as good as films on a new, fresh substrate.
A second substrate used is strontium titanate (SrTiO3). This substrate
has a very good lattice match (a=3.90A). This substrate is the substrate of
choice for the PAC films, because it usually produces higher quality films. Its
141
biggest disadvantage, namely a high dielectric constant, is of no concern to us,
because we are not interested in microwave applications.
The substrate of choice for the a-axis films was lanthanum aluminate
(LaA1O3). Its lattice match is not quite as good as in the case of the SrTiO3
(a =3.79 A), but considering that we were developing a new technique, we
needed a workhorse substrate again. In addition, the dielectric properties for
LaA103 are a lot better than for the SrTiO3, important in device applications,
which is one of the reasons a-axis films are interesting.
All substrates are subjected to the same cleaning procedures before
being loaded into the chamber. They are cleaned for 10 min. each in an
ultrasonic cleaner, first in acetone, then methanol, then ethyl alcohol. In
between each bath, they are carefully dabbed dry with a Kim Wipe, because air-
drying them leaves residues on the surface of the substrate.
A.2 Loading the Chamber
Loading the chamber is a top-down process. First the substrate is placed
in the substrate holder with the polished side down. The heater is positioned
on top of the holder and tested to make sure that is does not wiggle.
Because
the barium oxidizes rather rapidly, the
quartz-crystal
monitoring the barium boat has to be cleaned each time to remove any
oxidized and therefore loose barium. This is done by removing the crystal
holder, gently pushing the crystal out of the holder, cleaning the crystal with
alcohol and putting it back into the holder and the holder back into the
chamber. If this procedure has not been done before the barium boat was
outgassed, it has to be done now.
142
Otherwise the chamber can be loaded now. Barium pieces, which have
been carefully dabbed to remove most of the oil in which they were stored in
order to avoid or at least slow down oxidation, are put into the barium boat
and the yttrium and copper boat are loaded with the respective metals.
The last step is to check the shutter. Its function is to protect the
substrate from the evaporants while the rates still stabilize and to end the
deposition by covering the substrate before the boats are being switched off.
Testing the working of the shutter also gives a chance to check if the substrate
is still in the proper position after the heater has been placed on top of it.
A.3 Heating the Substrate before Deposition
After the chamber has been pumped down to 10-7 torr a film can be
made. First the substrate has to be preheated. This is done to get rid of
anything on the surface of the substrate, including water. The heating
procedure is a three step procedure, controlled by a computer program:
heating at 10°C/min to 400°C
staying at 400°C for 10 min
heating at 10°C/min to 600°C
staying at 600°C for 10 min
heating at 10°C/min to 720°C
staying at 720°C for 30 min
after the 30 min at 720°C the temperature is ramped manually to the desired
temperature for the deposition.
In addition, it is a good idea to switch on the quartz-rate monitors now.
One reason is that it gives the crystals time to stabilize. The other reason is
that if anything is wrong with the quartz-rate monitors, especially something
143
that requires opening the chamber, the chamber is not that hot yet and one
does not have to wait too long for it to cool down (there is not possibility to
actively cool the chamber) and therefore saves time.
A.4 Depositing the Film
Before the actual deposition is done, the choice has to be made about
which orientation the film is supposed to grow in. This influences the
substrate temperature during deposition and to a lesser degree the oxygen
pressure during the deposition. For a c-axis film on MgO, the substrate
temperature is 650 °C, while for the same kind of film on SrTiO3, a substrate
temperature of 670 °C is used. For the a-axis films on LaA1O3, the temperature
is reduced to between 580 °C and 600 °C.
In order to grow the right structure, it is important to supply some
oxygen during the deposition. For a "standard", orthorhombic, c-axis oriented
film, the pressure is set at 25% of 50 sccm, which yields a back-ground
pressure of about 5.5x10-5 torr, with the local pressure at the substrate being in
the mtorr range. For the a-axis films, in order to maximize the amount of
film growing in the right orientation, it is advantages to lower the pressure to
10% to 15% of 50 sccm, even though that reduces the transition temperature.
In addition, the shutter has to be switched on and set on open (it will
not actually open until the operate button is pressed). And most importantly,
the power to the boats has to be switched on and the settings on the power
controllers have to be checked. This is most important for the barium boat,
because two different settings are used for the outgassing and the actual
deposition.
144
After everything has been set to the desired values, the actual
deposition can be started. Similar to the heating of the substrate, the boats are
heated in a two step process. After the rates stabilized and the oxygen has been
switched on, the shutter is opened and the actual deposition is started. While
the deposition is going on, not much can be done by the operator except
noting if the rates are stable enough or if they are that unstable that the
deposition has to be stopped before the desired thickness has been reached.
Our films are deposited at 500A /min, which means that a 2000A thickness is
reached after 4 min. After the desired thickness has been reached, the
deposition is stopped by closing the shutter and then switching off the boats.
A.5 Annealing the Film
This step is responsible for the oxygen content of the film. The
experiments have shown that it is only feasible to achieve two different
oxygen contents by the in-situ anneal: one which is almost 7 and therefore
has the orthorhombic structure and an oxygen content of 6.25, which shows
the tetragonal structure. For an orthorhombic film, the chamber is backfilled
with oxygen to a pressure of 15 torr, the temperature is ramped to 450°C at
10°C/min and the film stays at that temperature for 30 min. For a tetragonal
film, the oxygen is switched off as fast as possible and the film is annealed i n
vacuum, which means that the pump pumps on the chamber for the whole
time of the anneal. The temperature during the anneal is 500°C. Higher
temperatures increase the amount of impurities formed. The anneal time is
usually around 1 h, with longer anneals again introducing more impurities.
145
A.6 Cool-down after Deposition
There are two basic options after the anneal. The film can either be
ramped down slowly at 20°C/min. This is used for the orthorhombic films in
order to make sure that the films stay fully oxygenated.
The other option is a quench, meaning that the film is cooled down as
fast as possible. We do not have the option here of doing a true quench by
dropping the film in liquid nitrogen, therefore our quench is done by
switching off the heater and letting the film and chamber cool down by itself.
As a summary, the temperature and to a lesser degree the oxygenpressure during the deposition are responsible for the orientation of the film,
while the structure of the film (orthorhombic or tetragonal) is controlled
mostly by the oxygen pressure during the anneal, with the temperature
mostly being responsible for the possible formation of impurities and
secondary phases.
146
Appendix B
Settings of X-Ray Diffractometer
Diffractometer: Siemens, D5000
Step drive: Normal, coupled
Step mode: continuous scan
Step time: 0.1s
Step size: 0.02°
Start position: 2°
Stop position: 120°
Divergence slit: 0.299°
Antiscatter slit: 0.299°
Rotation: None
Tube current: 30 mA
Tube voltage: 50 kV
147
Appendix C
Fitting of X-Ray Data
To determine the lattice parameters, the peak position are determine by
a fitting program called PROFILE, used in Dr. Sleight's lab. After choosing the
peaks to be fitted (the program can not fit the whole spectrum all at once, for
our film spectra, it is only able to fit one peak at a time), the program marks
the peaks to get a starting point for the fit. These peak positions can be
changed, if they do not seem likely. The raw data is fitted to the PseudoVoigtl function (but other fit functions are available)
+(ln)*2
1
=n
w)
(7.1)
1+(Xw-p)2
I is the intensity at the current angle, 10 the intensity at the peak position,
given by p in degrees 20, with X being the current angle and W=0.5*FWHM.
The parameters that are varied by this function to determine the best fit
are therefore Io, p, W and the mixing factor
because the Pseudo- Voigtl
function is a combination of a Lorentzian and a Gaussian. The quality of the
fit is determined by to quantities the reliability (Rel);
Rel =
'calf
W * as
*100%
(7.2)
w* Io2bs
and a theoretical reliability (TR), which takes into account the statistical
character of the observed intensities:
TR =
INV Iobs
*I
2
obs
y2
* 100%
(7.3)
148
W(x) is a weight function, usually set to give equal importance to all
data points, but can be adjusted to either give the peaks or the data near the
background more importance. The fit is accomplished by minimizing Rel and
the better the fit, the closer TR is to Rel. For peaks with a high number of
counts, Rel = 4-5%, but for small and therefore noisy peak, Rel can be as high
as 30%.